Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 08 Apr 2024 12:51:03 +0000
From:      bugzilla-noreply@freebsd.org
To:        bugs@FreeBSD.org
Subject:   [Bug 277992] mpr and possible trim issues
Message-ID:  <bug-277992-227-bebGreI8bn@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-277992-227@https.bugs.freebsd.org/bugzilla/>
References:  <bug-277992-227@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D277992

--- Comment #5 from mike@sentex.net ---
Created attachment 249831
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=3D249831&action=
=3Dedit
charting write throughput (90 second pauses between tests)

Good news / bad news. The test continued through the night without causing =
any
errors, however it is getting progressively slower.  From the disks being
fresh, it would take about 20min. Now the same test takes about 2hrs

I let it run overnight and each iteration got longer and longer
Apr  8 00:10:45 r-14mfitest LOOP[57086]: ending job ran (1:21:6)
Apr  8 00:12:16 r-14mfitest LOOP[57737]: starting
Apr  8 01:41:51 r-14mfitest LOOP[95255]: ending job ran (1:29:35)
Apr  8 01:43:21 r-14mfitest LOOP[95888]: starting
Apr  8 03:57:36 r-14mfitest LOOP[52511]: ending job ran (2:14:15)
Apr  8 03:59:06 r-14mfitest LOOP[53144]: starting
Apr  8 05:57:41 r-14mfitest LOOP[2987]: ending job ran (1:58:35)

the 3am one was probably an extra 15min due to various daily cron jobs runn=
ing.
But you can see the initial test that worked with decent write speed from t=
he
attached graph. Then the crash of the disk. I then disable TRIM and try aga=
in,
but its slower and slower

I have a feeling if I do a trim -f da[each disk]p1 one by one and resilver,=
 it
will then be fast again on writes for a bit

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-277992-227-bebGreI8bn>