Hi, colleagues.
In December 2017 I wrote to freebsd-net@ about tons of errors in log-file
Jan 5 09:49:40 server kernel: swap_pager_getswapspace(32): failed
Jan 5 09:49:40 server kernel: swap_pager_getswapspace(24): failed
Jan 5 09:49:40 server kernel: swap_pager_getswapspace(18): failed
This errors had happened after I add netgraph rule for my IPFW ruleset. This was
just ng_patch for setting TTL on OUTGOING interface.
/usr/sbin/ngctl -f- <<-SEQ
mkpeer ipfw: patch 100 in
name ipfw:100 ttl_set
msg ttl_set: setconfig { count=1 csum_flags=1 ops=[ \
{ mode=1 value=128 length=1 offset=8 } ] }
SEQ
/sbin/ipfw add 15002 netgraph 100 ip from me to any { recv re2 or recv epair* }
And I did not established the reason for this error, just netgraph rule was
removed.
After that, after playing with ipfw/fq_codel in VirtualBox I wanted setup it in
the production. But after some hours of the work, the old error was occurred.
I have 2 identical servers (ZFS mirror, VNET'ed Jails, IPFW ruleset and so
on). The difference is only in CPU, amount of RAM and connection to Internet.
This server uses pppoe via built-in ppp client.
The second server works fine without any errors.
I do not understand, is this problem is due to incorrect IPFW ruleset or
something else?
Without netgraph/dummynet rules I see in top -HSP a lot of processes of MySQL.
But innodb_buffer_pool_size is just 2GB. Server has 16GB RAM.
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
11570 106 20 0 570M 516M select 3 0:38 0.00% clamd{clamd}
11570 106 20 0 570M 516M select 1 0:00 0.00% clamd{clamd}
10612 88 20 0 2909M 451M uwait 3 0:07 0.01% mysqld{mysqld}
10612 88 20 0 2909M 451M uwait 3 0:04 0.01% mysqld{mysqld}
.....
uname -a
FreeBSD server 11.1-STABLE FreeBSD 11.1-STABLE #0 r325503: Tue Nov 7 13:38:44
EET 2017 wishmaster at servers:/usr/obj/usr/src/sys/MY amd64
Please help.
--
Vitaly