YunoHost-Apps / hedgedoc_ynh

HedgeDoc package for YunoHost
https://hedgedoc.org/
GNU General Public License v3.0
13 stars 8 forks source link

upgrade 1.9.6~ynh2 to 1.9.7~ynh2 is failing #60

Closed isAAAc closed 1 year ago

isAAAc commented 1 year ago

Describe the bug

upgrade to 1.9.7~ynh2 is failing

Context

Steps to reproduce

i expect an upgrade to 1.9.7~ynh2

Logs

i clean yarn cache and retry:

root@gourzoot:~# sudo yarn cache clean
yarn cache v1.22.19
success Cleared cache.
Done in 0.07s.
ericgaspar commented 1 year ago

no logs?

isAAAc commented 1 year ago

no logs?

i pressed enter to quickly ^^ the second try is on his way

isAAAc commented 1 year ago

humm, it failed again , but i saw something with the ram use. i stoped netdata (with eat some ram) , clean the yarn cache again, and restart a third time the upgrade:

2023-03-13 19:57:33,824 SUCCESS  yunohost.app (unknown function) - [1937983.1] Upgrade complete

so i think i need more ram on my server ^^

sorry for the noise, and thx for your ears !

i close the issue ;)

lapineige commented 1 year ago

I have a similar issue at about 93% too: Node enters Heap out of memory. Adding 10GB of swap (out of 1GB of free RAM, and using zram) doesn't solve it.

I tried this trick to increase the limit : https://stackoverflow.com/questions/38558989/node-js-heap-out-of-memory

It doesn't even go that high before crashing. Also at the same 93% level.

This software is using tremendous amount of memory just for a basic upgrade :(

lapineige commented 1 year ago

Ok, solved by adding 2 extras GB of RAM. It doesn't seem to care about swap space (at least no compressed RAM acting as swap such as Zram).

ericgaspar commented 1 year ago

How much ram did Hedgedoc used to install in your system?

lapineige commented 1 year ago

I wasn't able to monitor precisely. Somewhat less that 4GB I would say.

But the thing is that before I gave it a lot more than than of zram "swap", and it didn't use it. So either the memory usage is not possible to compress, or it quits when RAM is full, no matter how much swap is available.