* Theodore Ts'o (tytso@mit.edu) wrote:

On Mon, Aug 14, 2023 at 09:02:53PM +0000, Dr. David Alan Gilbert wrote:

quoted

dg         29594   29592  0 18:40 pts/0    00:00:00 /usr/bin/ar --plugin /usr/libexec/gcc/x86_64-redhat-linux/13/liblto_plugin.so -csrDT src/intel/perf/libintel_perf.a src/intel/perf/libintel_perf.a.p/meson-generated_.._intel_perf_metrics.c.o src/intel/perf/libintel_perf.a.p/intel_perf.c.o src/intel/perf/libintel_perf.a.p/intel_perf_query.c.o src/intel/perf/libintel_perf.a.p/intel_perf_mdapi.c.o

[root@dalek dg]# cat /proc/29594/stack 
[<0>] md_super_wait+0xa2/0xe0
[<0>] md_bitmap_unplug+0xd2/0x120
[<0>] flush_bio_list+0xf3/0x100 [raid1]
[<0>] raid1_unplug+0x3b/0xb0 [raid1]
[<0>] __blk_flush_plug+0xd7/0x150
[<0>] blk_finish_plug+0x29/0x40
[<0>] ext4_do_writepages+0x401/0xc90
[<0>] ext4_writepages+0xad/0x180

If you want a few seconds and try grabbing cat /proc/29594/stack
again, what does the stack trace stay consistent as above?

I'll get back to that and retry it.

Also, if you have iostat installed (usually part of the sysstat
package), does "iostat 1" show any I/O activity on the md device?
What about the underying block dvices used by the md device?  If the
md device is attached to HDD's where you can see the activity light,
can you see (or hear) any disk activity?

It's spinning rust, and I hear them go quiet when the hang happens.

Dave

This sure seems like either the I/O driver isn't processing requests,
or some kind of hang in the md layer....

				- Ted

-- 
 -----Open up your eyes, open up your mind, open up your code -------   
/ Dr. David Alan Gilbert    |       Running GNU/Linux       | Happy  \ 
\        dave @ treblig.org |                               | In Hex /
 \ _________________________|_____ http://www.treblig.org   |_______/