rspamd container stuck in restart loop after update.sh

hugalafutro

Hi, I ran the update.sh script today and afterwards the rspamd container is stuck in restart loop with re_cache; rspamd_re_cache_load_hyperscan: fatal error: cannot allocate scratch for /var/lib/rspamd//791648bcd295379bd35146fd029ee28c88cfa0bd2429269cf09abc9541528fc8.hs: -6

output of update.sh: https://o.o5.ddns.net/S5H8T
rspamd log: https://o.o5.ddns.net/jHhHJ

Any ideas how to make it work? Thanks!

hugalafutro

weirdly enough, it turns out it was because I also happened to change the server type some time before the update as I was just remembering what changes I made recently (hetzner cloud vm, rescaled form 2vcpu intel to 4vcpu amd)

Rescaling back to 2vcpu intel instance makes rspamd work.

Is there any way to make rspamd work on amd instance ?

accolon

Rspamd uses cache files that are CPU dependent. It looks like this even affects different x64 vendors like AMD and Intel.

See here:
https://rspamd.com/doc/faq.html#why-am-i-getting-errors-after-moving-rspamd-to-a-different-platform-cpu-type

hugalafutro

accolon thanks for the link that would explain it, guess I’ll stick with the 2vcpu, the next 4vcpu intel is twice the price.