<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Le 11/12/2018 à 10:28, Julien Rey a
écrit :<br>
</div>
<blockquote type="cite"
cite="mid:5C0F8329.1060009@univ-paris-diderot.fr">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<div class="moz-cite-prefix">Le 10/12/2018 13:33, <a
class="moz-txt-link-abbreviated"
href="mailto:quentin.bouget@cea.fr" moz-do-not-send="true">quentin.bouget@cea.fr</a>
a écrit :<br>
</div>
<blockquote cite="mid:5f32d4fd-c1b3-b4ed-c9be-36b022ea304a@cea.fr"
type="cite">
<div class="moz-cite-prefix">Le 10/12/2018 à 12:00, Julien Rey a
écrit :<br>
</div>
<blockquote type="cite"
cite="mid:5C0E4760.4030606@univ-paris-diderot.fr">Hello, <br>
<br>
We are running lustre
2.8.0-RC5--PRISTINE-2.6.32-573.12.1.el6_lustre.x86_64. <br>
<br>
Since thursday we are getting a "bad address" error when
trying to write on the lustre volume. <br>
<br>
Looking at the logs on the MDS, we are getting this kind of
messages : <br>
<br>
Dec 10 06:26:18 localhost kernel: Lustre:
9593:0:(llog_cat.c:93:llog_cat_new_log()) lustre-MDD0000:
there are no more free slots in catalog <br>
Dec 10 06:26:18 localhost kernel: Lustre:
9593:0:(llog_cat.c:93:llog_cat_new_log()) Skipped 45157
previous similar messages <br>
Dec 10 06:26:18 localhost kernel: LustreError:
9593:0:(mdd_dir.c:887:mdd_changelog_ns_store())
lustre-MDD0000: cannot store changelog record: type = 6, name
= 'PEPFOLD-00016_bestene1-mc-SC-min-grompp.log', t =
[0x20000a58f:0x858e:0x0], p = [0x20000a57d:0x17fd9:0x0]: rc =
-28 <br>
Dec 10 06:26:18 localhost kernel: LustreError:
9593:0:(mdd_dir.c:887:mdd_changelog_ns_store()) Skipped 45157
previous similar messages <br>
<br>
<br>
I saw here that this issue was supposed to be solved in 2.8.0:
<br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="https://jira.whamcloud.com/browse/LU-6556">https://jira.whamcloud.com/browse/LU-6556</a>
<br>
<br>
Could someone help us unlocking this situation ? <br>
<br>
Thanks. <br>
<br>
</blockquote>
<p>Hello,</p>
<p>The log messages don't point at a "bad address" issue but
rather at a "no space left on device" one ("rc = -28" -->
-ENOSPC).</p>
<p>You most likely have, at some point, registered a changelog
user on your mds and that user is not consuming changelogs.</p>
<p>You can check this by running:</p>
<pre>[mds0]# lctl get_param mdd.*.changelog_users
mdd.lustre-MDT0000.changelog_users=
current index: 3
ID index
cl1 0
</pre>
<p>The most important thing to look for is the distance between
"current index" and the index for "cl1", "cl2", ...<br>
I expect for at least one changelog user, that distance is
2^32 (the maximum number of changelog records).<br>
Note that changelog indexes wrap around (0, 1, 2, ...,
4294967295, 0, 1, ...).</p>
<p>If I am right, then you can either deregister the changelog
user:</p>
<pre>[mds0]# lctl --device lustre-MDT0000 changelog_deregister cl1
</pre>
<p>or acknowledge the records:</p>
<pre>[client]# lfs changelog_clear lustre-MDT0000 cl1 0
</pre>
<p>(clearing with index 0 is a shortcut for "acknowledge every
changelog records")</p>
<p>Both those options may take a while.</p>
<p>There is a third one that might yield faster result, but it
is also much more dangerous to use (you might want to check
with your support first) :<br>
</p>
<pre>[mds0]# umount /dev/mdt0
[mds0]# mount -t ldiskfs /dev/mdt0 /mnt/lustre-mdt0
[mds0]# rm /mnt/lustre-mdt0/changelog_catalog
[mds0]# rm /mnt/lustre-mdt0/changelog_users
[mds0]# umount /dev/mdt0
[mds0]# mount -t lustre /dev/mdt0 <...> # remount the mdt where it was
</pre>
<p><b>I cannot garantee this will not trash your filesystem. Use
at your own risk.<br>
</b></p>
<p>---</p>
<p>In recent versions (2.12, maybe even 2.10), lustre comes with
a builtin garbage collector for slow/inactive changelog users.</p>
<p>Regards,<br>
Quentin Bouget<br>
</p>
</blockquote>
<br>
Hello Quentin,<br>
<br>
Many thanks for your quick reply.<br>
<br>
This is what I got when I issued the command you suggested:<br>
<br>
<pre>[root@lustre-mds]# lctl get_param mdd.*.changelog_users</pre>
<pre>mdd.lustre-MDT0000.changelog_users=</pre>
<pre>current index: 4160462682</pre>
<pre>ID index</pre>
<pre>cl1 21020582</pre>
<br>
I then issued the following command:<br>
<pre>[root@lustre-mds]# lctl --device lustre-MDT0000 changelog_deregister cl1</pre>
<br>
It's been running for almost 20 hours now. Do you have an
estimation of the time it could take ?<br>
</blockquote>
When you deregister a changelog user: every changelog record has to
be invalidated (maybe this is batched, but I don't know enough about
the on-disk structure to say).<br>
<br>
I do not recall ever waiting that long. Then again, I never
personally deregistered a changelog users with that many pending
changelog records.<br>
<br>
If you just want to make sure Lustre is doing something, you can
have a look at your mdt0: invalidating changelog records should
generate a high load of small random writes.<br>
If the device is idle, something is probably wrong.<br>
<br>
Is your filesystem still unavailable?<br>
<br>
<blockquote type="cite"
cite="mid:5C0F8329.1060009@univ-paris-diderot.fr"> <br>
Best,<br>
<pre class="moz-signature" cols="72">--
Julien REY
Plate-forme RPBS
Molécules Thérapeutiques In Silico (MTi)
Université Paris Diderot - Paris VII
tel : 01 57 27 83 95 </pre>
<!--'"--><br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
lustre-discuss mailing list
<a class="moz-txt-link-abbreviated" href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a>
<a class="moz-txt-link-freetext" href="http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a>
</pre>
</blockquote>
Quentin<br>
</body>
</html>