<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<p>Hello everyone,</p>
<p>We have a Linstor cluster with HA managed by Pacemaker.</p>
<p>All DRBD resources are created as LVM thin volumes.<br>
</p>
<p>Like the <a moz-do-not-send="true"
href="https://linbit.com/drbd-user-guide/linstor-guide-1_0-en/#s-linstor_ha">documentation</a>
recommends, the Linstor database is inside a DRBD resource managed
by Linstor.</p>
<p>But something happened with the database resource, and it's now
impossible to enable it on another node, because it refuses to
become secondary, even when unmounted.<br>
</p>
<p>When trying to make it secondary with `drbdadm secondary` it
fails with the following error:</p>
<p># drbdadm secondary linstor_db<br>
linstor_db: State change failed: (-12) Device is held open by
someone<br>
additional info from kernel:<br>
/dev/drbd1000 opened by mount (pid 4007210) at 2022-02-27
17:49:02.233<br>
Command 'drbdsetup secondary linstor_db' terminated with exit code
11</p>
<p>The resource is not mounted, lsof and fuser come out empty
(expect for the drbd1000_submit process). The process 4007210 from
the error text doesn't exist anymore.</p>
<p># ls -l /dev/drbd/by-res/linstor_db/0 <br>
lrwxrwxrwx 1 root root 17 Mar 10 11:31
/dev/drbd/by-res/linstor_db/0 -> ../../../drbd1000<br>
</p>
<p># lsof 2>/dev/null | grep drbd1000<br>
drbd1000_ 4006238 root cwd DIR 253,1 4096 2 /<br>
drbd1000_ 4006238 root rtd DIR 253,12 4096 2 /<br>
drbd1000_ 4006238 root txt unknown
/proc/4006238/exe</p>
<p># ps -fp 4006238<br>
UID PID PPID C STIME TTY TIME CMD<br>
root 4006238 2 0 Feb27 ? 00:00:00
[drbd1000_submit]<br>
</p>
<p>Could anyone help me find a solution for this problem that
doesn't involve rebooting? Any help would be really appreciated!<br>
</p>
</body>
</html>