That’s what I said.
I think it’s not best practice generate every 5 minutes into file message which saying “everything is ok”. “Everything is OK” is default state and everyone expecting that, so they don’t have to be informed.
1 Like
I know 
What numbers are good and which aren’t ? That depends on the system and if you have issues it’s good to see it. Of course there maybe should be an option to disable it.
Hi,
I have isolated it a bit more for you, it seem that the “new” way of writing cache config in seahub_settings.py don’t work with 6.2 Pro or 6.2.3 CE, you need to write it in the old way.
Se in the bottom of this thread:
That’s not any good I guess as the new way to write is the right way to call memcached if I’m understand it correctly.
Old way that works in 6.2 Pro and 6.2.3:
> CACHES = { > 'default': { > 'BACKEND': 'django.core.cache.backends.memcached.MemcachedCache', > > # 'BACKEND': 'django_pylibmc.memcached.PyLibMCCache', > 'LOCATION': 'unix:/var/run/memcached.sock', > } > }
Do you have any errors in seahub_django_request.log that are related to memcache?
@daniel.pan any news on this?
I’ll clean everything up so it’s easier to understand.
Memcached with socket and HTTP are working in 6.2.2 without any issues.
When you update to 6.2.3 it start to log issues in seahub.log (Se log above)
This also happens in 6.2 Pro.
It seem that it only log errors when you write the Memcached settings in seahub_settings.py in the “new way” but if you write it in the “old way” everything seem to work.
To reproduce:
Have the new settings of memcached in seahub_Settings.py
Upgrade from 6.2.2 to 6.2.3.
Open your browser trough either NGINX or Apache2 and open a folder in the browser with thumbnails or something that triggers memecached.
Then you will have a error in seahub.log.
1 Like
After a few more test, we can reproduce this error.
But we found that this error msg ONLY showed in the first few HTTP requests after your rebooting Seahub, after you have used Seahub GUI for a little time, this error This error no longer appears any more.
Is that the case for you?
I don’t know if it disaper over time or not, I did only notice it directly and it did pop-up during my tests.
I can however run the server again for a test-run if you want me to do that over the weekend.
Edit:
But I do remember that after every reboot or restart the error did happen again.
I think we found what caused this problem.
Can you add this config to seafile-server-latest/runtime/seahub.conf
threads = 5
and test it again ?
In 6.2.2, there is such config there in seafile-server-latest/runtime/seahub.conf, but we removed it from 6.2.3.
4 Likes
Does it metter where in the file I add the line?
No, I think not.
It did work! I have tried it brefly and I did try to reproduce the issue but I can’t I’ll let 6.2.3 run a while now I’ll report back later.
1 Like
Happy to hear that.
2 Likes
Greate 
Then I guess the issue are the same at 6.2 Pro also.
@Cisco take a look in this thread you have the fix here for the Memcached issue Just five posts up.
2 Likes
Why did this line cause this issue? What’s that line for? And why did you delete that line?
I’m just curious =P
2 Likes
Pro version ? It looks like the Pro download area is still 6.1.9. Is 6.2x still considered early beta for Pro usage ?
1 Like
server-session.c(271): Unknown database type: pgsql
6.2.2 was ok with PostgreSQL
I also got the issue on Debian 9 Stretch and could solve it on the same way, thanks guys!
On 6.2.3 in comparison i’m now experiencing slow seahub performance for generating thumbs and loading the folder contents. Syncing via the Client runs smooth on all OS anyway in testing env as well as productive environment.
I’m going to test of disabling memcached changes that and update you.
After testing, the memcache error only occur after Seahub started for a few seconds. Later the error disappear. It is mostly likely a problem that memcache is not properly initialized after Seahub started. Apart from this, there are no other observable problems.
So I think this error is not an issue.
MemcachedCache is not thread-safe when Seahub is configured to use multi-thread mode. So we use PyLibMCCache.
I’m confused, the fix was to add threads = 5 according to @lian but did I understand you correct that we should not add that to the config?
2 Likes
Can I ask why PostgreSQL is not supported in 6.2.3?
server-session.c(271): Unknown database type: pgsql
6.2.2 was ok with PostgreSQL