Re: zfs with operations like rm -rf takes a very long time recently

From: void <void_at_f-m.fm>
Date: Sun, 16 Oct 2022 22:33:28 UTC 
On Sun, 16 Oct 2022, at 22:06, Warner Losh wrote:

> The disk is quite busy and performing decently well for the load...  
> The quite busy
> part I see others are trying to explain and/or diagnose, so I'll leave 
> them to it.

It runs ssl/nextcloud/postgresql13/nginx/memcached/exim but for only a few users. I'd not say it's particularly busy usually.

last pid:  5664;  load averages:  0.31,  0.24,  0.20                                                                                                 up 0+05:29:57  23:30:10
69 processes:  1 running, 68 sleeping
CPU:  0.7% user,  0.0% nice,  1.9% system,  0.1% interrupt, 97.4% idle
Mem: 391M Active, 237M Inact, 5304M Wired, 40K Buf, 1882M Free
ARC: 2623M Total, 1857M MFU, 149M MRU, 143K Anon, 79M Header, 537M Other
     1587M Compressed, 21G Uncompressed, 13.30:1 Ratio
Swap: 16G Total, 16G Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE    C   TIME    WCPU COMMAND
 4369 www           1  23    0   480M   148M accept   0   0:45   2.20% php-fpm
 1451 www           1  68    0   514M   151M accept   2   0:37   0.00% php-fpm
 5016 www           1  34    0   478M   145M accept   1   0:20   0.00% php-fpm
 1130 unbound       1  20    0   153M   119M kqread   0   0:10   0.00% unbound
 1292 root          3  20    0    19M  3320K select   1   0:09   0.00% apcupsd
 1345 postgres      1  23    0  2153M    74M select   3   0:05   0.00% postgres
 1351 postgres      1  20    0    31M    10M kqread   3   0:03   0.00% postgres
 1474 nobody       10  20    0    51M  7188K kqread   3   0:03   0.00% memcached