10:26:41 <ignatenkobrain> #startmeeting ssd_cache_test_day_f20
10:26:41 <zodbot> Meeting started Sun Oct 13 10:26:41 2013 UTC.  The chair is ignatenkobrain. Information about MeetBot at http://wiki.debian.org/MeetBot.
10:26:41 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
10:27:50 <ignatenkobrain> #topic SSD Cache Test Day: https://fedoraproject.org/wiki/Test_Day:2013-10-13_SSD_Cache
10:36:23 <k3olo> hi guys, i have my home system with hdd /dev/sdb and ssd /dev/sda. on ssd i have volume group vg_bb. vg_bb have about 3 gb of free space. I create lv vg_bb/bcache (2.5 gb).
10:37:22 <k3olo> can I use vg_bb/bcache instead of /dev/sdb1 in prerequisites?
10:40:51 <k3olo> rolffokkens: bon appetit =)
10:58:52 * adamw sent out blog post and forum news post about the test day
10:58:59 <adamw> may be too late to get people aware of it though :(
10:59:06 <adamw> i can't help testing sadly as i don't have the hardware around atm
11:06:41 <rolffokkens> adamw: you can also test in a VM. Not the performance of course, but you can test the other aspects.
11:07:22 <adamw> rolffokkens: hmm, i can try, not even sure i can squeeze sufficiently sized disks onto this laptop :)
11:08:16 <rolffokkens> k3olo: I did some testing with bcache on LV's. It worked, so you can try as well. But the test day pages have no docs on that.
11:09:22 <k3olo> rolffokkens: ok, i'll try and tell you about results
11:09:33 <rolffokkens> k3olo: thx
11:10:18 <rolffokkens> adamw: you need about 10 GB free space.
11:10:31 <adamw> rolffokkens: hum, might be able to scrape that together
11:15:56 <rolffokkens> I checked in my own VM: alle 2GB values in the test day page are the double of the bare minumum, so if you read 1 GB it'll still work.
11:17:02 <rolffokkens> my root FS needs actually 917MB.
11:17:23 <rolffokkens> adamw: so a total of 5GB ,ay be sufficient to test.
11:24:38 <rolffokkens> k3olo: you may also consider testing in a VM. virtual SSD is on vg_bb/bcache and virtual HDD somewhere on your HDD.
11:33:28 <ignatenkobrain> so. /me here
11:34:02 <ignatenkobrain> k3olo: at what step you now ?
11:38:23 <k3olo> another question - can bcache destroy my data in /home (/sdb3) ?
11:38:46 <ignatenkobrain> k3olo: destroy in casual use ?
11:39:29 <ignatenkobrain> if so that can. who knowns, but some critical issues was fixed
11:39:31 <k3olo> ignatenkobrain: during https://fedoraproject.org/wiki/QA:Testcase_bcache-tools_home_on_bcache_(no_LVM)#Setup or test
11:41:30 <ignatenkobrain> k3olo: yup. it will destroy all data on step "howto test"
11:42:31 <ignatenkobrain> .fire nirik
11:42:31 <zodbot> adamw fires nirik
11:42:33 <rolffokkens> k3olo: I guess "Setup, item 4" is not clear on this.
11:43:42 <k3olo> okay, you saved me )
11:43:45 <ignatenkobrain> k3olo: you need backup you /home/* files to another place and restore it after
11:44:24 <rolffokkens> Changed the wiki text. If needed "backup" all data in /home to another place on your root filesystem, because all data on /dev/sda2 will be destroyed.
11:44:24 <k3olo> ignatenkobrain: unfortunately I have not another place to store data
11:44:59 <k3olo> rolffokkens: and make it bold =)
11:45:41 <ignatenkobrain> k3olo: you can backup it to external device or to e.g. /var/tmp
11:45:43 <ignatenkobrain> or the same
11:46:25 <rolffokkens> k3olo: bold it is!
11:46:49 <rolffokkens> k3olo: I mean, changed it to bold :-)
11:47:09 <ignatenkobrain> rolffokkens: do for it 72pt font size and paint it to white :D
11:49:56 <rolffokkens> :-)
11:56:56 <rolffokkens> k3olo: at what step are you?
11:58:25 <k3olo> rolffokkens: going away because I can't take it. :(
11:58:35 <k3olo> *can't test
11:58:50 <rolffokkens> k3olo: what's the problem?
11:59:25 <ignatenkobrain> rolffokkens: I think not have space on / for backup
11:59:35 <k3olo> rolffokkens: the problem is I have not enough space to backup /home
11:59:44 <rolffokkens> You have 3 GB available on your SSD. How mucht space on your HDD?
11:59:56 <k3olo> (but I can test on mock partition on hdd)
12:00:10 <k3olo> rolffokkens: I have about 7 gb unallocated on hdd
12:00:17 <rolffokkens> k3olo: you may also consider testing in a VM. virtual SSD is on vg_bb/bcache and virtual HDD somewhere on your HDD.
12:00:18 <ignatenkobrain> k3olo: this ! test w/ mock
12:00:23 <ignatenkobrain> this good test
12:01:01 <ignatenkobrain> create mock partitions /var/cache/mock and /var/lib/mock on HDD
12:01:05 <ignatenkobrain> on bcache
12:01:23 <ignatenkobrain> and run build pkgs
12:01:32 <ignatenkobrain> rolffokkens: I think we could add this as test
12:02:01 <ignatenkobrain> rolffokkens: I think if our koji don't runs on ssd w/ bcache building packages should be faster
12:02:10 <ignatenkobrain> nirik: what do you think ?
12:03:03 <ignatenkobrain> too much time of building uses disk for installing dependencies nad et.c
12:03:08 <ignatenkobrain> s/nad/and/
12:03:24 <k3olo> ignatenkobrain: >and run build pkgs
12:03:24 <k3olo> ?
12:03:52 <k3olo> I don't know how to build something )
12:03:54 <rolffokkens> ignatenkobrain: you mean mock as a use case also for performance testing?
12:04:06 <rolffokkens> k3olo: how bout VM?
12:04:16 <ignatenkobrain> k3olo: it's easy!
12:04:18 <ignatenkobrain> $ fedpkg co gnome-shell
12:04:20 <ignatenkobrain> $ cd gnome-shell
12:04:22 <ignatenkobrain> $ fedpkg srpm
12:04:24 <ignatenkobrain> $ mock blahblahblah
12:04:29 <ignatenkobrain> rolffokkens: not almost
12:04:33 <k3olo> ignatenkobrain: ok
12:04:41 <ignatenkobrain> rolffokkens: not / or /home on bcache
12:04:46 <ignatenkobrain> mock partitions on bcache I mean
12:05:00 <ignatenkobrain> additional testcase
12:05:22 <ignatenkobrain> for this cases (when not enough space for backuping data)
12:05:30 <rolffokkens> k3olo: I'm not sure of testers all are familiar with mock.
12:05:56 <rolffokkens> k3olo: please respond: have you considered VM?
12:06:16 <k3olo> rolffokkens: no
12:06:28 <rolffokkens> k3olo: why not?
12:06:51 <rolffokkens> k3olo: no issues is losing data, ideal for testing
12:06:58 <rolffokkens> s/is/in/
12:07:31 <k3olo> rolffokkens: I have not enough time, I need to go after ~40 min
12:07:45 <k3olo> and I haven't f20 vm
12:08:04 <rolffokkens> ignatenkobrain: I'm not sure of testers all are familiar with mock
12:08:10 <k3olo> rolffokkens: but I can use my 8 gb of unallocated space on hdd to test.
12:08:34 <k3olo> if you tells me, how
12:08:36 <rolffokkens> k3olo: OK. Later you could do the test, there's enough space.
12:08:57 <rolffokkens> k3olo: ever used "virt-manager"?
12:09:09 <k3olo> rolffokkens: yes, only vbox
12:09:15 <ignatenkobrain> k3olo: oh no..
12:10:34 <k3olo> vbox have many nice feautures in user interface like easy device integration, shared clipboard, operations with screen of vm
12:10:39 <k3olo> ignatenkobrain: ^
12:11:00 <ignatenkobrain> rolffokkens: let me create testcases for mock partitions
12:11:14 <k3olo> ignatenkobrain: it is very user friendly, if you need gui in virtual machine
12:11:29 <ignatenkobrain> k3olo: no.
12:11:30 <ignatenkobrain> :P
12:12:04 <rolffokkens> ignatenkobrain: sure, could be interresting in real life to speedup koji.
12:13:26 <k3olo> bye, have a nice day
12:13:35 <ignatenkobrain> nirik: Kevin, сut it to sleep!
12:35:10 <ignatenkobrain> adamw: ping
12:55:33 <adamw> ignatenkobrain: ahoyho
12:55:34 <adamw> y
12:55:52 <ignatenkobrain> adamw: how much you know how runned our koji ?
12:56:11 <ignatenkobrain> adamw: we have genial idea
12:59:54 <adamw> what do you mean exactly?
13:01:44 <ignatenkobrain> adamw: we have genial idea to make another partition for cache and etc. for mock (koji use it) for speeding up koji in IOwait ;)
13:02:33 <adamw> ah, I don't know at that level of detail, no
13:02:44 <adamw> might be best to talk to nirik about that I think?
13:03:12 <ignatenkobrain> adamw: he's not available now. but generally what do you think about our idea?
13:03:44 <ignatenkobrain> I'll send mail to maillists today at the night about this too
13:08:30 <adamw> i really don't know how much of a bottleneck it is for koji or if something like this would help, sorry!
17:01:35 <roshi> how is the testing going?
17:01:57 <ignatenkobrain> roshi: not good :( one participant :(
17:02:06 <ignatenkobrain> you could be second
17:02:33 <roshi> I'm getting ready to spin up a VM right now :)
17:03:13 <roshi> part of the issue might be people don't use bcache - or know what it is
17:03:23 * roshi had to look it up
17:18:04 <rolffokkens> roshi: thanks for your feedback. I added a brief explanation on the Test page.
17:18:31 <roshi> sweet :)
17:22:57 <rolffokkens> roshi: are focussing on a particular test case?
17:23:10 <roshi> reading :)
17:23:24 <roshi> want to make sure I know what I'm setting up before I start
17:24:24 <rolffokkens> In case anything is unclear, there's a lot of support available for you right now! :-)
17:24:38 <roshi> probably start with 1.b because it hasn't been run yet
17:24:45 <roshi> true :)
17:26:54 <rolffokkens> FYI 1.b requires 1.a as a preparation.
17:27:20 <roshi> then I'll record my results and start there :)
17:29:20 <roshi> installing minimal now
17:31:35 <rolffokkens> roshi: taken care of the right partitioning?
17:32:08 <roshi> doing that now
17:34:24 <roshi> for a VM, I've added a second drive (I'm using virt-manager)
17:34:34 <roshi> how do I set it as an SSD
17:35:14 <roshi> ?
17:35:22 <rolffokkens> roshi: so you have a /dev/vda (sda) as HDD and a /dev/vdb (sdb) as SSD.
17:35:58 <roshi> I'm setting up the hw configuration now
17:35:59 <rolffokkens> roshi: if you have a real SSD you could create a disk image on that device.
17:36:17 <roshi> didn' t know if I could set it in the vm config
17:36:19 <roshi> I don't
17:36:19 <ignatenkobrain> roshi: rolffokkens, also. I don't know about virt-manager, but qemu can set speed for disks
17:36:20 <roshi> :(
17:36:37 <roshi> virt-manager is using qemu
17:36:44 <ignatenkobrain> and tests can be passed on one real hdd
17:36:56 <ignatenkobrain> but vda will set to 10mb/s
17:37:03 <ignatenkobrain> and vdb set to 100mbs
17:37:06 <ignatenkobrain> and test
17:37:09 <rolffokkens> roshi: then just create a disk. There are no options to make it behave like an SSD in virt-,amager
17:37:34 <rolffokkens> Except for performance there is no difference.
17:37:46 <roshi> I didn't think there was much difference
17:37:58 <roshi> but, having never used an SSD, I wasn't sure
17:38:11 <ignatenkobrain> for optional testing mock will do too much difference ;)
17:42:14 <rolffokkens> roshi: on the outside they look like any 2.5'' drive with the same SATA connectors, and to your system they're juist hard disks as well.
17:42:48 <roshi> I've worked with them - I just don't have any personally
17:45:27 <rolffokkens> I waited to buy one because I don't want to decide what data should be on the SSD and what data should be on the HDD. But when SSD caching became available, I bought one.
17:53:12 * roshi is running into space issues
17:54:17 <ignatenkobrain> roshi: what issues ?
17:54:36 <roshi> not enough disk space on host drive
17:54:42 * roshi fixing now
17:55:30 <roshi> virt-manager (or perhaps qemu, not sure) saves images in /var/lib/libvirt/images -> which was out of space
17:56:01 <ignatenkobrain> roshi: oh dude ;) delete you porn-vide^W^W^W yum cache :D
17:56:11 <ignatenkobrain> roshi: you can fix it in virt-manager
17:56:26 <roshi> I just got rid of some VM's I don't need anymore
17:56:43 * roshi needs to be more proactive with keeping files organized
17:57:12 <ignatenkobrain> you can use "Edit"->"Preferences"->"Storage"->"Add"
17:57:17 <ignatenkobrain> in virt-manager
17:57:19 <ignatenkobrain> ;)
17:59:26 <roshi> ep
17:59:32 <roshi> s/ep/yep
18:00:34 <roshi> wait, I have no 'storage' in preferences (not in the gui at least)
18:01:16 <ignatenkobrain> ah. I'm wrong
18:01:24 <ignatenkobrain> not Preferences. Connection details
18:01:58 <ignatenkobrain> roshi: http://storage9.static.itmages.com/i/13/1013/h_1381687317_7018361_d414f717c1.png
18:02:59 <roshi> I got it :)
18:14:52 <rolffokkens> roshi: creating VM is going well?
18:15:14 <roshi> from what I can tell
18:15:38 <roshi> handling the partitions now
18:18:56 <rolffokkens> roshi: In Anaconda I never actually made the partitions on the (virtual) SSD, allways did that later.
18:19:20 <ignatenkobrain> jreznik: hey ;)
18:19:21 <roshi> I'm doing it in anaconda right now
18:19:28 <rolffokkens> OK.
18:21:18 <roshi> we'll see if it works
18:21:23 <roshi> :)
18:22:14 <roshi> in the interest of being thorough - should we specify a swap size on the wiki for the partitioning?
18:22:33 * roshi is using 1GB for the minimal install
18:22:35 <ignatenkobrain> roshi: no. use needed swap size
18:22:52 <ignatenkobrain> use 1 GiB ;)
18:23:16 <ignatenkobrain> I don't like GBs. I'd like GiBs :D
18:23:59 <roshi> lol
18:26:54 * roshi is getting hungry
18:27:21 <ignatenkobrain> roshi: `date` ?
18:27:30 <roshi> ?
18:27:49 <roshi> 2013-10-13
18:27:54 <roshi> around lunch time
18:27:55 <ignatenkobrain> roshi: time ;)
18:27:58 <ignatenkobrain> I meant
18:28:02 <roshi> well, a bit past lunch time
18:28:08 <ignatenkobrain> ah
18:28:41 * roshi has the VM running - starting test cases
18:28:54 <ignatenkobrain> roshi: excellent!
18:31:10 <rolffokkens> roshi: great! I wasn't able to create all partition is F20 Anaconda.
18:31:48 <roshi> well, I think I got it all working ;)
18:31:53 <roshi> time will tell
18:32:05 <rolffokkens> roshi: or fdisk :-)
18:32:41 <ignatenkobrain> rolffokkens: gdisk for GPT (all UEFI systems)
18:33:49 <rolffokkens> ignatenkobrain: not in the VM. Well, not in my VM's.
18:34:30 <ignatenkobrain> rolffokkens: not in QEMU/KVM VMs, probably in vbox VMs and real machines
18:34:33 <ignatenkobrain> also
18:34:41 <ignatenkobrain> we should write EFI virtualiztion
18:36:24 <rolffokkens> https://bugzilla.redhat.com/show_bug.cgi?id=477035
18:36:38 <jreznik> ignatenkobrain: hi man!
18:37:27 <ignatenkobrain> rolffokkens: I should packaging OVMF
18:37:33 <ignatenkobrain> jreznik: how are you ?
18:37:41 <ignatenkobrain> jreznik: will you test ;)
18:37:43 <ignatenkobrain> ?
18:38:52 <rolffokkens> jreznik: yes, we need another tester! :-)
18:39:18 <ignatenkobrain> jreznik: we need mooore experimental hamsters :D
18:40:42 <jreznik> :D
18:40:51 <jreznik> how is it going?
18:41:40 <ignatenkobrain> jreznik: one human in table, second roshi lunching and testing, third human will test after 30-40 mins.
18:42:55 <ignatenkobrain> nothing more
18:43:11 <jreznik> well, it's pretty specific test day...
18:43:15 <ignatenkobrain> we need jreznik for testing ;)
18:48:35 <jreznik> I'd be killed at home if I won't shutdown laptop now :(
18:49:20 <ignatenkobrain> rolffokkens: I think for next testday (F21 ssd cache) we could update tests w/ testing performance on ssd+hdd real, hdd VM (ristrict IO for one of virtual disks)
18:49:34 <ignatenkobrain> jreznik: heh. sad
18:50:32 <rolffokkens> jreznik: bcache isn't worth getting killed
18:51:21 * rolffokkens wonder if that's actually understandable English.
18:54:16 <rolffokkens> ignatenkobrain: could be a nice test.
18:55:36 <roshi> part of the issue is this is a Sunday
18:55:47 <roshi> here in the states almost no one works on sunday
18:57:30 <rolffokkens> work as in "getting paid for" or "playing with computers"?
18:57:52 <roshi> lol
18:58:16 <roshi> sorry - was meaning that basically people hang out with friends/family, watch TV play games
18:58:26 <roshi> even nerds usually have something to be doing
18:58:38 <roshi> if I had the time I would be out camping right now :) or hiking
18:58:50 * roshi has been trying to find some time to go fishing...
18:59:05 <ignatenkobrain> roshi, reform USA ;)
19:00:39 <rolffokkens> interesting that in the US watching TV is considered a better way to spend time than playing with computers :-)
19:01:03 <rolffokkens> Unless you mean watching TV with friends/familiy.
19:01:38 <ignatenkobrain> but why not playing on piano/trombone or the same ? :(
19:02:07 <roshi> it's a US thing
19:02:18 * roshi doesn't have TV
19:02:34 <roshi> too expensive :p
19:02:47 <roshi> would rather spend money on good beer and food :)
19:03:09 <ignatenkobrain> in Russia we name TV as zombie box
19:04:10 <rolffokkens> beer and food: primary needs in life.
19:04:15 <rolffokkens> and internet.
19:04:28 <ignatenkobrain> s/beer/vodka/g
19:04:30 <ignatenkobrain> fixed
19:05:17 <rolffokkens> I've seen a lot of zombie movies on TV, but that's not what you mean I guess :-)
19:05:39 * roshi is a fan of vodka
19:05:57 <ignatenkobrain> roshi: zombie box because TV is zombiing people
19:05:58 <roshi> I use the TV for games and an extra monitor when I really need it
19:06:10 <roshi> I follow :)
19:07:09 <ignatenkobrain> roshi: redhat looking for you, alcoholic! :D
19:07:20 <ignatenkobrain> roshi: how about tests ?
19:07:25 <rolffokkens> roshi: lunching or also torturing your VM with test cases?
19:07:55 <rolffokkens> hm, the support people are a little eager
19:08:07 <roshi> figuring out what to eat for lunch
19:08:11 <roshi> :p
19:08:24 <roshi> also, double checking my partitions
19:08:38 * roshi usually lets the installer handle the partitioning
19:09:15 * rolffokkens wonders if and roshi really did that
19:09:40 <roshi> ?
19:10:06 <rolffokkens> s/if and/if and how/
19:10:57 <roshi> did what?
19:11:10 <rolffokkens> F20 anaconda only wants to make a partition when I assign a filesystem to it. But for the (virtual) SSD that's too early.
19:11:31 <roshi> yeah - I don't think it worked like I thought it did :)
19:11:42 <roshi> hence, me rechecking the partitions
19:12:03 <rolffokkens> Hm. what does fdisk /dev/vda show?
19:12:25 <ignatenkobrain> rolffokkens: fdisk -l might be more beautiful
19:12:58 * rolffokkens agrees
19:14:08 <ignatenkobrain> almost 9h was passed :D
19:14:10 <roshi> it seems my /swap got put on the ssd instead of the hdd
19:14:46 <ignatenkobrain> roshi: but you should have free space on ssd
19:14:52 <roshi> I do
19:14:54 <rolffokkens> that's no problem - you can do without swap during the test.
19:14:55 <ignatenkobrain> and you can use it for test
19:15:25 <rolffokkens> And the partition order on sda/vda is OK?
19:16:12 <roshi> not sure
19:16:39 <ignatenkobrain> roshi: # fdisk -l | fpaste
19:16:43 <ignatenkobrain> roshi: provide please
19:18:08 <roshi> https://www.irccloud.com/#!/ircs://irc.freenode.net:6697/%23fedora-test-day
19:18:31 <roshi> meh - pentadactyl pasting....
19:18:36 <roshi> /me grumbles
19:18:51 <roshi> http://paste.fedoraproject.org/46507/38169184/
19:20:09 <rolffokkens> Looks OK. is /dev/vda3 your root FS?
19:20:51 <rolffokkens> df may be an easy way to show that.
19:21:17 <ignatenkobrain> rolffokkens: too ;)
19:21:38 <ignatenkobrain> we should integrate df and fdisk to systemd :D we not needed dupe dunctional :D
19:21:43 <ignatenkobrain> s/d/f/
19:22:34 <roshi> http://ur1.ca/fvtk0
19:23:33 <roshi> vda2
19:23:49 <rolffokkens> The prereqs say that you need to have vda3 as root FS.
19:24:02 * rolffokkens thinks if we can work around this
19:24:44 <roshi> well, I did try to set this up via anaconda
19:25:04 <ignatenkobrain> I think all good
19:25:21 <ignatenkobrain> and not needed to reinstall
19:25:37 <rolffokkens> In Anacondo the order is really important. I first create /boot, then /home and then /
19:25:39 <ignatenkobrain> only replace 2 and 3 in commands
19:26:09 <roshi> ah - didn't know it was specific like that
19:26:34 <rolffokkens> yes, but for test case 1.b it'll be impossible to use the whole disk for your root FS.
19:26:38 <ignatenkobrain> rolffokkens: I think different configs more better to test
19:28:23 <rolffokkens> OK, /dev/sda2 could be used in the end as swap on bcache.
19:28:23 <roshi> well, I can reinstall - it doesn't take long with the minimal
19:28:38 <ignatenkobrain> roshi: no. don't do this
19:28:46 * roshi is getting ready to go grab a burrito :)
19:28:54 <roshi> I haven't
19:29:59 <rolffokkens> roshi: no need to do this. But plz remove swap from /etc/fstab and the virtual SSD.
19:30:31 <rolffokkens> roshi: if we do that, we'll be able to use /dev/vda2 as a bcached swap in the and.
19:30:44 <rolffokkens> s/we/you/
19:30:48 <ignatenkobrain> reinstall - shindows way
19:31:18 <roshi> removed swap from fstab
19:31:33 <ignatenkobrain> roshi: swapoff blahblah
19:31:46 <rolffokkens> swapoff -a
19:32:18 <roshi> done - brb
19:32:22 <roshi> er, bbiab
19:32:35 <roshi> (be back in a bit)
19:34:50 <rolffokkens> (back in a bite) :-)
19:36:29 <ignatenkobrain> rolffokkens: bite or bitte ? ;)
19:38:20 <ignatenkobrain> roshi is burrito :D
19:40:01 <ignatenkobrain> k3olo: how to go you testing ?
19:40:33 <k3olo> ignatenkobrain: I'm on the "create a filesystem: mkfs -t ext4 -L HOME /dev/bcache0" step =)
19:40:37 <nirik> ignatenkobrain: FYI, caching in koji is off so we have reproducable builds. We currently know every package version thats in every buildroot used by every build. If there was caching it would be a lot more complex. You're welcome to try and talk to dgillmore and see if there's some way to do it that still makes things reproducable.
19:41:00 <k3olo> ignatenkobrain: but i'll mount it into my ~/bcache_test dir
19:41:11 <ignatenkobrain> nirik is alive ;) wow
19:41:24 <nirik> ignatenkobrain: I was on vacation, just got back.
19:42:08 <ignatenkobrain> nirik: I've pinged dgilmore for take a look for my critical bug, but have no pong ;(
19:42:17 <nirik> what critical bug?
19:42:46 <ignatenkobrain> nirik: in GitPython
19:42:46 <ignatenkobrain> searching
19:42:52 <ignatenkobrain> also I've requested acls for it
19:42:58 * k3olo reboot
19:43:13 <nirik> ok, just curious. I have no connection to GitPython...
19:43:25 <ignatenkobrain> nirik: https://bugzilla.redhat.com/show_bug.cgi?id=1010706
19:43:55 <ignatenkobrain> nirik: I'm using it in my kernel-package program
19:44:55 <nirik> ok... I don't know what that is... but good luck. ;)
19:45:12 <k3olo> ignatenkobrain: I have not write access into ~/bcache_test
19:45:34 <ignatenkobrain> k3olo: from root ?
19:47:00 <ignatenkobrain> k3olo: so you need to `$ cd ~/bcache_test && sudo mkdir test && sudo chown blahblah:blahblah -R test` the same command
19:47:35 <ignatenkobrain> nirik: you can take a look for my pkg for qa :D
19:47:44 <ignatenkobrain> nirik: https://github.com/ignatenkobrain/kernel-package
19:49:36 <nirik> ignatenkobrain: I don't really have time to, but good luck with it.
19:49:56 <ignatenkobrain> nirik: thanks ;)
19:54:28 <k3olo> ignatenkobrain: не писал мне ничего?
19:54:44 <ignatenkobrain> k3olo: oh no. not a russian :D
19:54:56 <ignatenkobrain> k3olo: so you need to `$ cd ~/bcache_test && sudo mkdir test && sudo chown blahblah:blahblah -R test` the same command
19:55:52 <k3olo> ignatenkobrain: holy fuck, now they reveal us =)
19:58:28 <k3olo> ignatenkobrain: during reboot there was some freeze:
19:58:29 <k3olo> окт 13 23:47:04 bb.lan systemd[1]: sddm.service stopping timed out. Killing.
19:58:29 <k3olo> окт 13 23:47:04 bb.lan systemd[1]: sddm.service: main process exited, code=killed, status=9/KILL
19:58:29 <k3olo> окт 13 23:47:04 bb.lan systemd[1]: Stopped Simple Desktop Display Manager.
19:58:29 <k3olo> окт 13 23:47:04 bb.lan systemd[1]: Unit sddm.service entered failed state.
19:59:33 <k3olo> i didn't see it before
20:01:11 <ignatenkobrain> k3olo: you mounted bcache as /home ?
20:01:28 <k3olo> ignatenkobrain: no, as /home/nedr/bcache_test
20:01:43 <ignatenkobrain> k3olo: ok. so. now is ok ?
20:02:05 <ignatenkobrain> k3olo: mount | fpaste
20:02:39 <k3olo> ignatenkobrain: yes, all is ok. how can i test io speed?
20:02:54 <k3olo> *everything is ok
20:03:20 <rolffokkens> k3olo: excellent.
20:04:18 <k3olo> rolffokkens: I want to test speed, any ideas about that?
20:04:20 <ignatenkobrain> k3olo: # dd if=/dev/zero of=/home/user/bcache_test/test.file bs=64M and see iotop
20:04:29 <k3olo> ok
20:04:42 <rolffokkens> k3olo: or copy (a lot of) files from and to the bcache device.
20:05:11 <ignatenkobrain> py1hon: are you here?
20:05:58 <rolffokkens> k3olo: the dd suggestion may not work really well, because bcache tends to not use the cache for sequential I/O.
20:06:28 <ignatenkobrain> rolffokkens: probably
20:06:40 <k3olo> rolffokkens: how can I see I/O speed while copying files?
20:06:49 <ignatenkobrain> k3olo: iotop
20:06:53 <ignatenkobrain> or iostat
20:07:07 <rolffokkens> k3olo: you may also need to change your cache_mode to writeback, default is writethrough
20:07:30 <k3olo> rolffokkens: where can I change it?
20:07:35 <rolffokkens> writethrough works only for read.
20:07:52 <rolffokkens> k3olo: first check please. what is the cache_mode?
20:08:03 <rolffokkens> use bcache-status to see
20:08:50 <k3olo> rolffokkens: writethrough
20:09:24 <rolffokkens> OK. What's your backing device for /dev/bcache0?
20:09:40 <k3olo> rolffokkens: /dev/sdb3
20:10:49 <rolffokkens> k3olo: echo writeback > /sys/block/sdb/sdb3/bcache/cache_mode
20:12:07 <k3olo> rolffokkens: Cache Mode                  writethrough [writeback] writearound none
20:12:17 <rolffokkens> right!
20:13:53 <rolffokkens> "iostat 1 sda sdb" will show you the I/O
20:15:14 <k3olo> rolffokkens: hm, I see in KDE's transfers: copying begins from 30 MB/s but now it stops at all at ~2 GB done
20:16:20 <rolffokkens> k3olo: you mean the copy command is blocked?
20:17:06 <k3olo> rolffokkens: copy is just freezed without terminating
20:17:22 <rolffokkens> k3olo: sounds like an issue I read about....
20:17:50 <ignatenkobrain> .fire py1hon
20:17:50 <zodbot> adamw fires py1hon
20:18:11 <ignatenkobrain> py1hon: we need you ;)
20:19:07 * rolffokkens is searching...
20:24:24 <rolffokkens> http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2049
20:24:32 <rolffokkens> Looks like that.
20:27:50 <ignatenkobrain> rolffokkens: I can build kernel w/ this patch
20:27:51 <rolffokkens> k3olo: did you read: http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2049 ?
20:28:26 <k3olo> rolffokkens: reading it now
20:29:25 <rolffokkens> ignatenkobrain: building the kernel will take a while. For me there's 1.5 hour left, then to bed.
20:29:39 <k3olo> rolffokkens: looks like my case
20:29:41 <ignatenkobrain> rolffokkens: 25-30mins for me
20:30:07 <rolffokkens> ignatenkobrain: OK. plz try.
20:30:15 <ignatenkobrain> k3olo: can you wait ?
20:30:49 <rolffokkens> ignatenkobrain, k3olo: or later this week?
20:31:42 <k3olo> ignatenkobrain: no, I need to wake up in 6 hours
20:31:59 <k3olo> I can do it tomorrow
20:32:04 <ignatenkobrain> k3olo: ok. please fill a bug in kernel component
20:32:11 <ignatenkobrain> and CC us
20:32:38 <ignatenkobrain> Unfortunately py1hon busy today :9
20:32:42 <ignatenkobrain> it's very sad
20:32:45 <k3olo> ignatenkobrain: I can comment http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2049
20:32:49 <k3olo> with my dmesg
20:33:10 <ignatenkobrain> k3olo: more better to have bug for building and testing patches
20:33:18 <ignatenkobrain> https://bugzilla.redhat.com/
20:34:04 <rolffokkens> k3olo: py1hon is reading linux-bcache mail list actively....
20:34:50 <ignatenkobrain> rolffokkens: for adding patches as downstream we need to have bug at RHBZ ;)
20:35:57 <rolffokkens> ignatenkobrain: OK, but I'll get py1hon in the loop for a final solution
20:44:56 <k3olo> rolffokkens: https://bugzilla.redhat.com/show_bug.cgi?id=1018615
20:46:12 <ignatenkobrain> k3olo: cat /proc/cmdline
20:49:30 <k3olo> ignatenkobrain: cat /proc/cmdline
20:49:30 <k3olo> BOOT_IMAGE=/vmlinuz-3.11.4-301.fc20.x86_64 root=/dev/mapper/vg_bb-root ro rd.lvm.lv=vg_bb/root vconsole.font=latarcyrheb-sun16 vconsole.keymap=us rhgb quiet LANG=ru_RU.UTF-8
20:49:43 <k3olo> good night, rolffokkens, ignatenkobrain
20:50:01 <rolffokkens> k3olo: good night to ypu
20:50:24 <ignatenkobrain> k3olo: you forget write to table you results ;) https://fedoraproject.org/wiki/Test_Day:2013-10-13_SSD_Cache#Test_Results
20:51:41 <ignatenkobrain> rolffokkens: unfortunately you link isn't correctly for k3olo bug :(
20:56:18 <rolffokkens> why?
20:56:43 <ignatenkobrain> rolffokkens:  Igor Gnatenko 2013-10-13 16:49:19 EDT
20:56:47 <ignatenkobrain> Bug present in the kernel bcache realisation.
20:56:51 <ignatenkobrain> http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2049
20:56:55 <ignatenkobrain> but in 3.11.4 patch was applied:
20:56:57 <ignatenkobrain> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/md/bcache/request.c?id=c0f04d88e46d14de51f4baebb6efafb7d59e9f96
20:57:01 <ignatenkobrain> Rolf or I will send message to upstream bcache mailing list.
20:57:41 <roshi> 1.A finished without a hitch
20:57:55 <roshi> though, getting permissions errors when logging in with a test user
20:58:02 <ignatenkobrain> roshi: oh my god. you is still alive ! ;)
20:58:08 <roshi> can't cd to /home/<username>
20:58:10 <roshi> :D
20:58:19 <ignatenkobrain> roshi: ll /home/ | fpaste
20:58:28 <roshi> it takes more than some partitioning and a burrito to kill me
20:58:29 <rolffokkens> ignatenkobrain: could you please do in response to the mail I just tent to the list?
20:58:32 <roshi> ;)
20:59:04 <ignatenkobrain> rolffokkens: sure. will do. but tomorrow
20:59:23 <ignatenkobrain> roshi: provide paste of my command
20:59:50 <roshi> http://ur1.ca/fvtzy
20:59:56 <roshi> I was :p
21:00:14 <ignatenkobrain> roshi: sure. chown -R test:test /home/test
21:00:16 <rolffokkens> roshi: ownership issue
21:00:17 <ignatenkobrain> as root
21:00:23 <rolffokkens> yep
21:00:34 <roshi> right :)
21:02:24 <ignatenkobrain> also
21:02:26 <ignatenkobrain> $ stat -c %a /home/brain
21:02:28 <ignatenkobrain> 710
21:02:50 <ignatenkobrain> you need set chmod 0710 /home/test
21:04:22 <rolffokkens> roshi: so now case 1.B?
21:04:43 <roshi> that's the plan rolffokkens
21:05:15 <ignatenkobrain> roshi: how bout permissions ? fxd ?
21:05:57 <roshi> well, I fixed the permissions so that I can do things in the right home directory
21:06:12 <roshi> but for some reason permission is denied when I log in
21:06:27 <roshi> so, it doesn't drop the user in their home directory on login
21:06:46 <ignatenkobrain> probably $ stat -c %a /home/brain
21:06:48 <ignatenkobrain> 710
21:06:55 <ignatenkobrain> probably you don't have 710
21:07:04 <ignatenkobrain> and DM could block
21:07:27 <roshi> 710 is done
21:08:18 <rolffokkens> roshi: ll -d /home
21:11:26 <rolffokkens> roshi: case 1.B should be no problem, only the optional part at the end does not apply.
21:12:12 * roshi thinks it could have something to do with selinux policies
21:12:28 <rolffokkens> roshi: that's is very well possible!
21:12:51 <rolffokkens> roshi: please edit /etc/selinux/config
21:12:51 <roshi> setting it to permissive to see if that changes anything
21:13:30 <rolffokkens> roshi: selinux is something I should warn about in the test docs.
21:14:27 <roshi> that fixed it
21:14:50 <roshi> prompt still looks weird though: -bash-4.2$
21:15:05 <ignatenkobrain> oops
21:16:44 <roshi> well, first test passed - and I'm learning about bcache, so it's a good day
21:16:54 <roshi> for me, anyway
21:18:33 <rolffokkens> roshi: case 1.B should be no problem, only the optional part at the end does not apply.
21:19:08 <roshi> sounds good - I'll let you know if I have any hiccups
21:28:32 <ignatenkobrain> rolffokkens: http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2113
21:28:46 <ignatenkobrain> I will build custom kernel w/ this patch
21:30:14 <rolffokkens> ignatenkobrain: for k3olo?
21:30:21 <ignatenkobrain> rolffokkens: yep
21:30:36 <ignatenkobrain> rolffokkens:  I think it should fix hes problem
21:30:50 <rolffokkens> was that his problem? NULL ptr deref?
21:31:11 <ignatenkobrain> rolffokkens: I think so. I looked dmesg
21:31:19 <ignatenkobrain> [ 1550.135293] BUG: unable to handle kernel NULL pointer dereference at
21:31:21 <ignatenkobrain> 000000000000006c
21:31:29 <rolffokkens> OK.
21:37:47 <ignatenkobrain> roshi: how to go you ?
21:38:06 <roshi> good, 1.B is done
21:38:22 <roshi> everything went well - no errors and / is on bcache0
21:38:31 <ignatenkobrain> roshi: so good
21:38:35 <roshi> yeah
21:38:43 * roshi updates the matrices on wiki
21:38:58 <ignatenkobrain> roshi: also we need to test one bug
21:39:03 <ignatenkobrain> and probably fix
21:39:19 <ignatenkobrain> rolffokkens: could you tell to roshi how-to reproduce bug ?
21:41:51 * roshi goes for a quick run
21:42:41 <ignatenkobrain> rolffokkens 12:10:49 AM
21:42:44 <ignatenkobrain> k3olo: echo writeback > /sys/block/sdb/sdb3/bcache/cache_mode
21:42:53 <ignatenkobrain> roshi: replace sdb3 to you device
21:43:00 <ignatenkobrain> and copy files
21:48:14 <ignatenkobrain> roshi: actually you should get unreachable copy
21:48:18 <ignatenkobrain> i.e. bug
21:49:00 <rolffokkens> Was away for a while. roshi went smooth...
21:50:17 <ignatenkobrain> rolffokkens: we need to test bug which reported by k3olo and if it is still persist - test kernel, to which I referenced in the bug
21:51:04 <rolffokkens> ignatenkobrain: I can try to crash my own VM
21:51:34 <ignatenkobrain> rolffokkens: good. please test. I'll be back after 15-40mins
21:52:03 <rolffokkens> ignatenkobrain: about to go to bed - tomorrow monday.
21:52:41 <ignatenkobrain> rolffokkens: I'm not needed more sleep. I should not wake up early ;)
21:53:20 <ignatenkobrain> rolffokkens: but if you needs sleep - don't worry. we can continue tomorrow ;)
21:55:03 <rolffokkens> ignatenkobrain: doing some tests now
22:01:07 <rolffokkens> ignatenkobrain: interesting issue with VM: the underlying host OS (BTRFS) complains: BTRFS error (device dm-0): csum failed ino 3026069 off 1449168896 csum 3343383093 private 844597172
22:01:29 <rolffokkens> An the VM itself reports an I/O error.
22:02:13 <rolffokkens> The underlying OS is using BTRFS on bcache. Don;t know which one to blame.
22:02:36 <rolffokkens> Well, I'm off. See y'all later!
22:22:39 <roshi> ignatenkobrain: was there instructions on how to recreate that bug?
22:33:46 <ignatenkobrain> so
22:33:49 * ignatenkobrain is here
22:33:59 <ignatenkobrain> roshi: at what step are you ?
22:35:10 <ignatenkobrain> rolffokkens, I think btrfs is bad choose
22:43:49 <roshi> just got back from a quick jog
22:43:57 <roshi> I've finished 1.B
22:44:20 <ignatenkobrain> tested bug ?
22:44:37 <roshi> not yet
22:44:41 <ignatenkobrain> echo writeback > /sys/block/sdb/sdb3/bcache/cache_mode
22:44:54 <ignatenkobrain> replace sdb and sdb3 w/ you partition
22:45:35 <ignatenkobrain> IIRC for you case sdb == sda and sdb3 == sda2 or sda1
22:46:06 <roshi> `echo writeback`?
22:46:40 <ignatenkobrain> `echo writeback > ........`
22:48:27 <roshi> ok, done
22:48:42 <ignatenkobrain> roshi: copy (not move) data above folders
22:50:29 <roshi> copy data from anywhere to vda3?
22:50:44 <roshi> or just somewhere in /home
22:50:53 <roshi> is on bcache atm
22:51:03 <ignatenkobrain> to /home which on bcache
22:52:18 <roshi> kk
22:52:44 <ignatenkobrain> roshi: if it will stay on some % - you have this bug
22:55:53 <roshi> cp my_file.txt /home
22:55:55 <roshi> from /
22:55:57 <roshi> no error
22:56:25 <ignatenkobrain> big file
22:56:33 <roshi> kk
22:56:48 <ignatenkobrain> also. you can generate it w/ dd from /dev/urandom ;)
22:59:19 <roshi> yeah
22:59:33 <roshi> getting a big file now - netinst iso :)
22:59:53 <ignatenkobrain> i think it is nit a big
23:00:09 <ignatenkobrain> we copied >2GiB and on 2GiB it was frozen
23:00:50 <roshi> so you think I need bigger?
23:01:00 <ignatenkobrain> roshi: probably..
23:01:43 <roshi> hm
23:01:49 <roshi> wget for the iso dies
23:02:00 <ignatenkobrain> heh
23:02:05 <ignatenkobrain> don't panic
23:02:28 <roshi> I rarely panic :)
23:02:36 <ignatenkobrain> wait
23:03:10 <ignatenkobrain> press sysrq-t
23:03:48 <roshi> huh, systemd is having issues: user@1000.service entered a failed state
23:03:52 <roshi> sysrq?
23:04:08 <ignatenkobrain> SysRq or PrtScr
23:04:32 <roshi> ok
23:04:33 <roshi> why?
23:04:41 <ignatenkobrain> what why ?
23:04:49 <roshi> PrtScr - t
23:04:52 <roshi> why press that?
23:05:34 <ignatenkobrain> we need to get trace
23:05:35 <ignatenkobrain> http://thread.gmane.org/gmane.linux.kernel.bcache.devel/2049
23:05:37 <ignatenkobrain> argh
23:05:39 <ignatenkobrain> 't'     - Will dump a list of current tasks and their information to your
23:05:41 <ignatenkobrain> console.
23:05:57 <ignatenkobrain> after do - see dmesg
23:06:01 <ignatenkobrain> and paste it
23:06:42 <roshi> hm
23:06:58 <ignatenkobrain> damn. I'm wront
23:07:00 <ignatenkobrain> wrong
23:07:14 <ignatenkobrain> Ctrl - Alt - SysRq - t
23:07:15 <roshi> wget was doing 1.2 m/s down for a while - now it's down to 10kb/s
23:07:25 <ignatenkobrain> I know
23:08:01 <ignatenkobrain> do C-A-SysRq-t and provide dmesg | fpaste
23:08:03 <roshi> doing that on the host won't do anything
23:08:24 <ignatenkobrain> heh
23:08:40 <ignatenkobrain> if will do, but will not critical
23:09:05 <ignatenkobrain> but I hope it will do consider VM
23:09:16 <ignatenkobrain> damn. my brain
23:09:28 <ignatenkobrain> inside. not consider
23:10:59 <ignatenkobrain> .fire roshi
23:10:59 <zodbot> adamw fires roshi
23:11:27 <roshi> http://ur1.ca/fvum1
23:11:51 <ignatenkobrain> yup
23:11:53 <ignatenkobrain> this
23:11:55 <ignatenkobrain> you have bug
23:12:38 <ignatenkobrain> https://bugzilla.redhat.com/show_bug.cgi?id=1018615
23:12:46 <roshi> what line indicates the bug?
23:13:02 * roshi doesn't really know what he's testing at this point
23:13:17 <ignatenkobrain> roshi: se 568 string
23:13:28 <ignatenkobrain> unable to handle kernel NULL pointer dereference at
23:13:47 <roshi> ah
23:13:50 <ignatenkobrain> and you wget stopped or too slow. right  ?
23:13:58 <roshi> I stopped wget
23:14:07 <roshi> 10kb/s is a big slow
23:14:11 <ignatenkobrain> yup
23:14:16 <ignatenkobrain> please test new kernel
23:14:19 <ignatenkobrain> from koji
23:14:25 * ignatenkobrain looking for link
23:14:40 <ignatenkobrain> and sure CC to bug
23:14:47 <ignatenkobrain> http://koji.fedoraproject.org/koji/buildinfo?buildID=470953
23:19:59 <ignatenkobrain> roshi: at what step?
23:20:27 <roshi> installing rpm dev tools to upgrade the kernel
23:20:45 <ignatenkobrain> roshi: wxy you need rpm dev tools ? O_o
23:20:51 * roshi thinks that's what he needs
23:21:08 <ignatenkobrain> roshi: you are weird :D
23:21:10 <roshi> this is deeper stuff than I'm used to doing
23:21:18 * roshi is learning
23:21:27 <ignatenkobrain> # yum install http://kojipkgs.fedoraproject.org//packages/kernel/3.12.0/0.rc4.git4.1.fc21/x86_64/kernel-3.12.0-0.rc4.git4.1.fc21.x86_64.rpm
23:21:30 <ignatenkobrain> and etc
23:21:41 * roshi didn't think it would be that easy :p
23:22:41 <ignatenkobrain> u need kernel kernel-modules-extra and probably other pkg
23:22:43 <ignatenkobrain> s
23:23:38 <roshi> it keeps hanging on yum install
23:24:04 * roshi would guess that the 10kb/s is happening again
23:24:10 <ignatenkobrain> damn
23:24:32 <ignatenkobrain> have you fallback w/o/ bcache on root fs ?
23:24:47 <roshi> nope
23:25:18 <ignatenkobrain> ok. so.
23:25:35 <roshi> should I give yum a chance to do it's thing?
23:26:05 <ignatenkobrain> I think this not needed. it will freeze after sometime
23:26:31 <ignatenkobrain> update matrices in wiki w/ link on this bug and tomorrow we can quick re-test this. ok ?
23:27:02 <ignatenkobrain> wiki already was referenced to this bug
23:27:06 <ignatenkobrain> you should see
23:27:14 <roshi> which test is this regarding?
23:27:34 <roshi> I'll be on tomorrow - so I can take another crack at it
23:27:35 <ignatenkobrain> all ;)
23:27:50 <roshi> ah
23:28:03 <ignatenkobrain> it's general issue in kernel
23:28:09 <ignatenkobrain> not /home or / specific
23:29:01 <roshi> so update both test cases then
23:29:40 <roshi> also: thanks for the help, I learned a bit today :)
23:29:54 <ignatenkobrain> will do.
23:30:08 <ignatenkobrain> roshi: thanks for the testing ;)
23:30:28 <roshi> :)
23:30:37 <roshi> someone's got to do it :p
23:30:48 * roshi will know more about the process next time
23:34:33 <ignatenkobrain> updated testcases
23:35:41 <ignatenkobrain> roshi: thanks againg. and who tested thanks too.
23:35:48 <ignatenkobrain> should have a some sleep
23:35:56 <ignatenkobrain> all: good night!
23:36:11 <ignatenkobrain> record will be available
23:36:19 <ignatenkobrain> #endmeeting