How to determine why uwsgi is crashing?

Question

Django

Hello dear experts.
Once a month, my VPS steadily crashes and the logs show the lack of memory caused by python.
About 3 months ago it all worked in debug mode (manage.py runserver .......) and these problems started after I moved everything to uwsgi in emperor-mode.
The stable error in the logs is as follows:

kern.log

Jan  9 15:45:05 vps04104 kernel: [1578651.353004] Out of memory: kill process 25286 (zabbix_agentd) score 125810 or a child
Jan  9 15:45:05 vps04104 kernel: [1578651.353004] Killed process 25288 (zabbix_agentd)
Jan  9 15:45:05 vps04104 kernel: [1578651.434161] python invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0
Jan  9 15:45:05 vps04104 kernel: [1578651.435489] python cpuset=/ mems_allowed=0
Jan  9 15:45:05 vps04104 kernel: [1578651.436829] Pid: 24440, comm: python Not tainted 2.6.32-5-amd64 #1
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Call Trace:
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810b6174>] ? oom_kill_process+0x7f/0x23f
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810b6698>] ? __out_of_memory+0x12a/0x141
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810b67ef>] ? out_of_memory+0x140/0x172
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810ba54d>] ? __alloc_pages_nodemask+0x4e5/0x5f4
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810bbab1>] ? __do_page_cache_readahead+0x9b/0x1b4
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810bbbe6>] ? ra_submit+0x1c/0x20
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810b48c2>] ? filemap_fault+0x17d/0x2f6
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810ca82a>] ? __do_fault+0x54/0x3c3
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff8100c241>] ? __raw_callee_save_xen_pud_val+0x11/0x1e
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff810ccb53>] ? handle_mm_fault+0x3b8/0x80f
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff812fe876>] ? do_page_fault+0x2e0/0x2fc
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  [<ffffffff812fc715>] ? page_fault+0x25/0x30
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Mem-Info:
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA per-cpu:
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] CPU    0: hi:    0, btch:   1 usd:   0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] CPU    1: hi:    0, btch:   1 usd:   0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA32 per-cpu:
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] CPU    0: hi:  186, btch:  31 usd:  78
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] CPU    1: hi:  186, btch:  31 usd:  48
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] active_anon:363776 inactive_anon:121622 isolated_anon:0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  active_file:35 inactive_file:8 isolated_file:0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  unevictable:1107 dirty:0 writeback:0 unstable:0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  free:3434 slab_reclaimable:1855 slab_unreclaimable:3582
Jan  9 15:45:05 vps04104 kernel: [1578651.437064]  mapped:748 shmem:127 pagetables:10113 bounce:0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA free:8032kB min:32kB low:40kB high:48kB active_anon:1820kB inactive_anon:1948kB active_file:16kB inactive_file:0kB unevictable:0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] lowmem_reserve[]: 0 2004 2004 2004
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA32 free:5704kB min:5708kB low:7132kB high:8560kB active_anon:1453284kB inactive_anon:484540kB active_file:0kB inactive_file:32kB
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] lowmem_reserve[]: 0 0 0 0
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA: 2*4kB 9*8kB 3*16kB 5*32kB 3*64kB 1*128kB 3*256kB 3*512kB 3*1024kB 1*2048kB 0*4096kB = 8032kB
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Node 0 DMA32: 673*4kB 30*8kB 2*16kB 1*32kB 1*64kB 1*128kB 0*256kB 1*512kB 0*1024kB 1*2048kB 0*4096kB = 5748kB
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 6881 total pagecache pages
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 6066 pages in swap cache
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Swap cache stats: add 293524, delete 287458, find 7490/10319
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Free swap  = 0kB
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Total swap = 1048568kB
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 524288 pages RAM
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 11000 pages reserved
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 67652 pages shared
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] 507702 pages non-shared
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Out of memory: kill process 25286 (zabbix_agentd) score 112178 or a child
Jan  9 15:45:05 vps04104 kernel: [1578651.437064] Killed process 25289 (zabbix_agentd)
Jan  9 15:45:05 vps04104 kernel: [1578651.508537] uwsgi invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0
Jan  9 15:45:05 vps04104 kernel: [1578651.509423] uwsgi cpuset=/ mems_allowed=0
Jan  9 15:45:05 vps04104 kernel: [1578651.510357] Pid: 4468, comm: uwsgi Not tainted 2.6.32-5-amd64 #1

I start with this line:
/usr/local/bin/uwsgi --emperor /opt/config --emperor-nofollow --daemonize=/var/log/uwsgi.log The
vassal config looks like this:

[uwsgi]
uid = ................
gid = www
umask = 012
chdir = /opt/apps/%n/app
virtualenv = /opt/apps/%n/venv
master = true
threads = 2
socket = /var/run/uwsgi/%n.sock
logto = /opt/apps/%n/logs/uwsgi.log
processes = %k
harakiri = 120
max-requests = 200
env = DJANGO_SETTINGS_MODULE=app.settings
module = django.core.handlers.wsgi:WSGIHandler()

Please tell me which way to debug to find the error. Thanks

Answer 1

No, the problem seems to be found. The bottom line is that on the VPSk there was 2GB of RAM and in the muscle configs they were configured for a 4GB RAM machine, i.e. the maximum amount of RAM that the muscle ate was somewhere around 3.5GB, respectively, everything fell down.