[kwlug-disc] ZFS pool says it's 100% full according df -h and yet I know there is space as I have remove a lot of files...
Federer Fanatic
nafdef at gmail.com
Sun Dec 29 12:07:41 EST 2019
sudo systemctl status sanoid.service
● sanoid.service - Snapshot ZFS Pool
Loaded: loaded (/lib/systemd/system/sanoid.service; static; vendor
preset: enabled)
Active: inactive (dead) since Sun 2019-12-29 12:00:03 EST; 5min ago
Process: 10396 ExecStart=/usr/sbin/sanoid --take-snapshots --verbose
(code=exited, status=0/SUCCESS)
Main PID: 10396 (code=exited, status=0/SUCCESS)
Dec 29 12:00:01 XXXXX sanoid[10396]: INFO: taking snapshots...
Dec 29 12:00:01 XXXXX sanoid[10396]: taking snapshot
homedirs at autosnap_2019-12-29_12:00:01_daily
Dec 29 12:00:02 XXXXX sanoid[10396]: cannot create snapshots : out of space
Dec 29 12:00:02 XXXXX sanoid[10396]: CRITICAL ERROR: /sbin/zfs snapshot
homedirs at autosnap_2019-12-29_12:00:01_daily fai
Dec 29 12:00:02 XXXXX sanoid[10396]: taking snapshot
homedirs at autosnap_2019-12-29_12:00:01_hourly
Dec 29 12:00:03 XXXXX sanoid[10396]: cannot create snapshots : out of space
Dec 29 12:00:03 XXXXX sanoid[10396]: CRITICAL ERROR: /sbin/zfs snapshot
homedirs at autosnap_2019-12-29_12:00:01_hourly fa
Dec 29 12:00:03 XXXXX sanoid[10396]: INFO: cache expired - updating from
zfs list.
Dec 29 12:00:03 XXXXX systemd[1]: sanoid.service: Succeeded.
Dec 29 12:00:03 XXXXX systemd[1]: Started Snapshot ZFS Pool.
--------------------------------------------------------------
Roger Federer Fanatic Extraordinaire :-)
On Sun, Dec 29, 2019 at 11:54 AM Benjamin Tompkins <bjtompkins at gmail.com>
wrote:
> So this is actually expected. Once you get past 80% full on a ZFS pool,
> write performance starts to suffer. I don't have an article handy
> regarding this at the moment as I am on my phone.
>
> Once some space gets freed you should get performance back.
>
> On Sun., Dec. 29, 2019, 10:28 a.m. Federer Fanatic, <nafdef at gmail.com>
> wrote:
>
>> Seems to be the same problem, no space on device:
>>
>> sudo zdb -M homedirs
>> vdev 0 metaslabs 116 fragmentation 27%
>> 12: 185132 ****************************************
>> 13: 113286 *************************
>> 14: 72563 ****************
>> 15: 67800 ***************
>> 16: 59987 *************
>> 17: 30729 *******
>> 18: 13720 ***
>> 19: 7163 **
>> 20: 4157 *
>> 21: 1977 *
>> 22: 964 *
>> 23: 502 *
>> 24: 254 *
>> 25: 88 *
>> 26: 5 *
>> 27: 3 *
>> 28: 1 *
>> pool homedirs fragmentation 27%
>> 12: 185132 ****************************************
>> 13: 113286 *************************
>> 14: 72563 ****************
>> 15: 67800 ***************
>> 16: 59987 *************
>> 17: 30729 *******
>> 18: 13720 ***
>> 19: 7163 **
>> 20: 4157 *
>> 21: 1977 *
>> 22: 964 *
>> 23: 502 *
>> 24: 254 *
>> 25: 88 *
>> 26: 5 *
>> 27: 3 *
>> 28: 1 *
>>
>>
>>
>> --------------------------------------------------------------
>> Roger Federer Fanatic Extraordinaire :-)
>>
>>
>> On Sun, Dec 29, 2019 at 9:09 AM Federer Fanatic <nafdef at gmail.com> wrote:
>>
>>> Ok. I think I get what's going on. Available space is being released,
>>> painfully slowly ,
>>>
>>> zfs get all is show very active processing albeit very slow...
>>>
>>>
>>>
>>> --------------------------------------------------------------
>>> Roger Federer Fanatic Extraordinaire :-)
>>>
>>>
>>> On Sun, Dec 29, 2019 at 8:35 AM Federer Fanatic <nafdef at gmail.com>
>>> wrote:
>>>
>>>> Looks like some fragmentation is occurring...presumably because disk
>>>> was getting full, I read
>>>> that having dedup setting on can cause some issues, but my
>>>> zfs get all |grep dedup
>>>> homedirs dedup
>>>> off default
>>>>
>>>>
>>>>
>>>>
>>>> --------------------------------------------------------------
>>>> Roger Federer Fanatic Extraordinaire :-)
>>>>
>>>>
>>>> On Sun, Dec 29, 2019 at 8:23 AM Federer Fanatic <nafdef at gmail.com>
>>>> wrote:
>>>>
>>>>> Also zfs get all has a constant stream of output...
>>>>>
>>>>>
>>>>>
>>>>> --------------------------------------------------------------
>>>>> Roger Federer Fanatic Extraordinaire :-)
>>>>>
>>>>>
>>>>> On Sun, Dec 29, 2019 at 8:19 AM Federer Fanatic <nafdef at gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Specifically, zpool status
>>>>>>
>>>>>>
>>>>>> pool: homedirs
>>>>>> state: ONLINE
>>>>>> scan: scrub repaired 0B in 0 days 03:57:48 with 0 errors on Sun Dec
>>>>>> 8 04:21:49 2019
>>>>>> config:
>>>>>>
>>>>>>
>>>>>>
>>>>>> zpool status yields:
>>>>>>
>>>>>> NAME STATE READ WRITE CKSUM
>>>>>> homedirs ONLINE 0 0 0
>>>>>> mirror-0 ONLINE 0 0 0
>>>>>> sda ONLINE 0 0 0
>>>>>> sdb ONLINE 0 0 0
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> # zpool list homedirs
>>>>>>
>>>>>> NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP
>>>>>> HEALTH ALTROOT
>>>>>> homedirs 1.81T 1.76T 57.7G -
>>>>>> - 27% 96% 1.00x ONLINE -
>>>>>>
>>>>>> which show 1.76 TB. It looks like space is being locked down somehow
>>>>>> and is unavailable and yet
>>>>>>
>>>>>> and yet
>>>>>>
>>>>>> # df -h
>>>>>> homedirs 890G 890G 0 100%
>>>>>> /homedirs
>>>>>>
>>>>>> It's as if the volume is shrinking as a I remove files...
>>>>>>
>>>>>> NOTE. It may have something to do with snapshots sanoid settings:
>>>>>> as I am also seeing the mount
>>>>>>
>>>>>> Filesystem
>>>>>> Size Used Avail Use% Mounted on
>>>>>> homedirs at autosnap_2019-12-26_05:00:22_hourly 1.7T 1.7T 0
>>>>>> 100% /homedirs/.zfs/snapshot/autosnap_2019-12-26_05:00:22_hourly
>>>>>>
>>>>>> ---this goes away if I run systemctl stop sanoid.timer however,
>>>>>> df -h still yes same output as above
>>>>>>
>>>>>>
>>>>>> NOTE. the 1.7TB
>>>>>>
>>>>>> Also removing files from ZFS takes an eternity.
>>>>>>
>>>>>> I am obviously not understanding something.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> --------------------------------------------------------------
>>>>>> Roger Federer Fanatic Extraordinaire :-)
>>>>>>
>>>>> _______________________________________________
>> kwlug-disc mailing list
>> kwlug-disc at kwlug.org
>> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
>>
> _______________________________________________
> kwlug-disc mailing list
> kwlug-disc at kwlug.org
> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://kwlug.org/pipermail/kwlug-disc_kwlug.org/attachments/20191229/92f437e7/attachment.htm>
More information about the kwlug-disc
mailing list