Seems like my server may be affected by the 9.1.7 update, lots of weird problems

[scooke]

And the browser window would all be black. Macos, Chrome Version 147.0.7727.138 (Official Build) (x86_64)

[nebulon]

receiving response timeout from the backend usually means the box process is too slow to respond, which in turn may hint at a system overload. But since one is also failing when querying the fileystem routes, maybe you have a locked up filessytem mounted which makes the system hang?

Does a normal df -h via SSH work fine and are all mountpoints healthy?

[scooke]

Seems like it. I have lots of the below:

overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/42dc09e1ab2f10ed432cae69adfe96136fd88c3a605b4bc39f823c05cc53b679/merged
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/3ac8dfcbe628dcdf82236861f36b799f4392830bad9203669741cb2d9f1a0767/merged
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/74626746866ed86941b6e48831acf5f38a2feaa8297c3c4ad6851d4ec591c169/merged
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/1c7f0e1688f8c536f6d6081af1f98b8ba7033658f32a173e0eed1f41d62ae847/merged
tmpfs                                      4.8G  4.0K  4.8G   1% /run/user/1000
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/4a208722f4e76888c9ff9b9a9a69c1180f28458b075b8bdbc2a81f17029acbeb/merged
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/091f8c6db63e863e5046a8ab27351f49f6eef980cadda12dd5d4e06acbbfa491/merged
overlay                                    709G  281G  392G  42% /var/lib/docker/overlay2/358bb3005eb75b4de6a19ac978ca6e9acffaf301e587c3c65278c383ca6588a8/merged

I do have two remote mountpoints which seem fine, but one is at 95%.

[scooke]

I tried one app which was Not Responding, a github pages app, and it eventually started. I tried another, Syncthing, and it failed:

y 11 09:55:09 taskworker: Starting task 26861. Logs are at /home/yellowtent/platformdata/logs/d1047eec-0fb9-4f77-9e9a-befd6280935d/apptask.log
May 11 09:55:11 taskworker: Running task of type app
May 11 09:55:12 apptask: run: startTask installationState: pending_restart runState: running
May 11 09:55:12 tasks: updating task 26861 with: {"percent":10,"message":"Starting app services"}
May 11 09:55:12 tasks: updating task 26861 with: {"percent":30,"message":"Restarting container"}
May 11 09:55:28 tasks: updating task 26861 with: {"percent":70,"message":"Configuring reverse proxy"}
May 11 09:55:28 shell: openssl: openssl x509 -noout -subject -issuer
May 11 09:55:29 reverseproxy: providerMatches: subject=CN = sync.example.com domain=sync.example.com issuer=C = US, O = Let's Encrypt, CN = E8 wildcard=false/false prod=true/true issuerMismatch=false wildcardMismatch=false match=true
May 11 09:55:29 shell: openssl: openssl x509 -startdate -enddate -subject -noout
May 11 09:55:29 openssl: expiryDate: subject=CN = sync.example.com notBefore=May  2 08:28:20 2026 GMT notAfter=Jul 31 08:28:19 2026 GMT daysLeft=80.77279122685185
May 11 09:55:29 reverseproxy: ensureCertificate: sync.example.com acme cert exists and is up to date
May 11 09:55:29 reverseproxy: needsRenewal: false. ARI {"start":"Tue, 30 Jun 2026 11:12:24 GMT","end":"Thu, 02 Jul 2026 06:23:14 GMT","rt":"Wed, 01 Jul 2026 19:28:54 GMT","valid":"Mon, 11 May 2026 16:35:58 GMT","url":"https://acme-v02.api.letsencrypt.org/acme/renewal-info","ts":"Mon, 11 May 2026 09:28:27 GMT"}
May 11 09:55:30 reverseproxy: writeAppLocationNginxConfig: writing config for "sync.example.com" to /home/yellowtent/platformdata/nginx/applications/d1047eec-0fb9-4f77-9e9a-befd6280935d/sync.example.com.conf with options {"sourceDir":"/home/yellowtent/box","vhost":"sync.example.com","hasIPv6":true,"ip":"172.18.18.120","port":8000,"endpoint":"app","redirectTo":null,"certFilePath":"/home/yellowtent/platformdata/nginx/cert/sync.example.com.cert","keyFilePath":"/home/yellowtent/platformdata/nginx/cert/sync.example.com.key","robotsTxtQuoted":null,"cspQuoted":null,"hideHeaders":[],"proxyAuth":{"enabled":false,"id":"d1047eec-0fb9-4f77-9e9a-befd6280935d","location":"/"},"upstreamUri":"","hstsPreload":false}
May 11 09:55:30 shell: reverseproxy: /usr/bin/sudo --non-interactive /home/yellowtent/box/src/scripts/restartservice.sh nginx
May 11 09:55:39 tasks: updating task 26861 with: {"percent":100,"message":"Done"}
May 11 09:55:41 tasks: setCompleted - 26861: {"result":null,"error":null,"percent":100}
May 11 09:55:41 tasks: updating task 26861 with: {"completed":true,"result":null,"error":null,"percent":100}
May 11 09:55:41 taskworker: Task took 32.599 seconds
May 11 09:55:41 Exiting with code 0
May 11 09:55:47 => Healthcheck error: Error: connect ECONNREFUSED 172.18.18.120:8000
May 11 09:55:58 => Healthcheck error: Error: connect ECONNREFUSED 172.18.18.120:8000

[nebulon]

That last log reads like things have worked out but it is still waiting for the healthcheck to succeed, the actual task exited fine with status code 0.

Not sure now what we should look at and what the issue is? Is the system overall showing high load or so?

[scooke]

Running ps aux --sort=-%mem | head shows me that yellowt+ 115187 65.5 0.6 10317968 328268 ? Ssl 02:48 479:12 node /home/yellowtent/box/box.js is using 65.5% of the CPU... is that normal?
I also checked for CPU usage:

shane@my:~$ ps aux --sort=-%cpu | head
USER         PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
www-data  458271 74.2  0.2 225084 131340 ?       R    15:07   0:20 php /app/pkg/wp --path=/app/data/public/ cron event run --due-now
yellowt+  115187 65.6  0.6 10323796 334180 ?     Rsl  02:48 485:34 node /home/yellowtent/box/box.js
shane     457394 64.8  2.4 2466264 1207968 pts/0 Rl   15:05   1:34 [celeryd: celery@b7c9aacf-af6b-4def-9981-7f24a752ed7c:ForkPoolWorker-112]
root         922 33.1  0.2 8575356 128632 ?      Ssl  May10 312:52 /usr/bin/dockerd -H fd:// --log-driver=journald --exec-opt native.cgroupdriver=cgroupfs --storage-driver=overlay2 --experimental --ip6tables --userland-proxy=false
message+  458377 33.0  0.0 1304600 21680 ?       Ss   15:07   0:00 postgres: user29397f7ab06c47eb898f8db21d530154 db29397f7ab06c47eb898f8db21d530154 fd00:c107:d509::1(54210) idle
shane      46768 31.4  0.5 22570552 256684 pts/0 Rl   00:11 281:31 node current/index.js
mysql        908 30.4  0.6 3079376 336300 ?      Ssl  May10 287:19 /usr/sbin/mysqld
message+   27167 24.8  1.2 5731612 594168 ?      Sl   May10 231:05 /usr/sbin/mysqld
shane      29722 15.6  0.4 22525668 206996 pts/0 Sl   May10 145:30 node current/index.js

[nebulon]

If it is normal or not, kinda depends on what the app does and how much traffic it sees, so maybe you have some idea from the past experience on that server?

So to come back to the original issue, the main thing is, that some dashboard routes are timing out, causing 500 status codes on requests still?

[scooke]

@nebulon Yes, still timeouts. Just a few minutes ago the dashboard showing my apps loaded... AND SAID THERE WERE NO APPS INSTALLED!

[nebulon]

yeah so I guess the server struggles to respond in time (there is a time cap of 30s to respond to the dashboard). So the apps are still installed, but the dashboard failed to probably load them. Not sure how to debug this, is it possible that the box service just keeps restarting?

is systemctl status box showing some decent uptime of that process and if not, if you keep trailing the logs tail -f /home/yellowtent/platformdata/logs/box.log does it show frequent restarts?

Also is anything unusual happening in say journalctl --system -f ?

[scooke]

Well, I made a ticket at the host (SSDNodes) and they responded saying they "made some node-level tweaks", and now the server is back to its flying speed. I couched it in terms of any relation to two recent kernel local-privilege-escalation bugs (Copy Fail, CVE-2026-31431; Dirty Frag, CVE-2026-43284 plus CVE-2026-43500), but they didn't specify what they did. But, it is back to nice and very fast.

[nebulon]

glad it worked out in the end, when too many things fail all at once it is often a hardware/virtualization issue