-
Problem report
-
Resolution: Won't fix
-
Major
-
None
-
4.0.1
-
None
-
Server:
Debian 9 Stretch 4.9.0-7-amd64
Intel(R) Xeon(R) CPU E5-2630 v3 @ 2.40GHz
RAM 128G
DB:
Debian 9 Stretch 4.9.0-7-amd64
Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz
RAM 64G
1TB RAID10 LVM
Hi guys!
We have a problem with history syncers. Our setup seems ok, history syncer processes operating strangely. They don't sendto many insert's to db (see screenshots).
Steps to reproduce:
- 2.5K hosts on 8K NPVS installation
- 4 db syncer processes
- ps -C zabbix_server f |grep syncer
- strace -p PID1, PID2, PID3, PID4
Result:
sendto(5, "\203\0\0\0\3select clock,ns,value from "..., 135, 0, NULL, 0) = 135
sendto(5, "\351\1\0\0\3update triggers set state=0"..., 493, 0, NULL, 0) = 493
sendto(5, "j\0\0\0\3update problem set r_eventi"..., 110, 0, NULL, 0) = 110
sendto(5, "D\0\0\0\3select eventid,escalationid"..., 72, 0, NULL, 0) = 72
sendto(5, "Rv\0\0\3insert ... is rare
internal queue is all red
Expected:
many sql inserts (with items data) in sendto system calls
sendto(5, "Rv\0\0\3insert into history_uint (i"..., 8192, 0, NULL, 0) = 8192
sendto(5, "Rv\0\0\3insert ...
sendto(5, "Rv\0\0\3insert ...
sendto(5, "Rv\0\0\3insert ...
sendto(5, "Rv\0\0\3insert ...
OR am i missing smth?
ps is russian ok?