Very common problem
Seafile/Seahub can't start after upgrade or any other reasons
Please check whether the old version of Seahub is still running.
Please check whether you use the right user to run or upgrade Seafile. Pay special attention to the following files:
You can run the following command to change fix the permission for the whole directory:
chown -R userx:groupx seafiledirectory
You can also try remove the cache directory of Seahub
rm -rf /tmp/seahub_cache
Failed to upload/download file online
- Make sure your firewall for Seafile fileserver is opened.
- Make sure
SERVICE_URLin ccnet.conf and
FILE_SERVER_ROOTin seahub_settings.py are set correctly. Furthermore check that you haven't overwritten them using the settings in the Seahub Admin section.
- Use Chrome/Firefox debug mode to find out which address is being used when clicking download button and whether it is correct.
Seahub/Seafile started correctly, but when visiting the web interface, it shows "Internal Server Error"
It is mostly likely some required Python packages of Seahub is not installed correctly.
You can check the detailed error messages in
/var/log/nginx/seahub.error.log if you use Nginx.
Website displays "Page unavailable", what can I do?
You can check the back trace in Seahub log files (
You can also turn on debug mode by adding
DEBUG = Trueto
seahub_settings.pyand restarting Seahub with
./seahub.sh restart, then refresh the page, all the debug infomations will be displayed. Make sure ./seahub.sh was started as: ./seahub.sh start-fastcgi in case you're using fastcgi.
Failed to send email, what can I do?
Please check logs/seahub.log.
There are some common mistakes:
- Check whether there are typos in the config (
seahub_settings.py, e.g. you could have forgotten to add a single quote
EMAIL_HOST_USER = XXX, which should be
EMAIL_HOST_USER = 'XXX'or you could have a space at the end of a config line.
- Your mail server is not available.
Can't connect to LDAP server with ldaps
Seafile server can't communication with my LDAP server. The ccnet.log shows:
[08/05/16 09:47:17] ../common/session.c(398): Accepted a local client [08/05/16 09:47:17] user-mgr.c(335): ldap_initialize failed: Bad parameter to an ldap routine. [08/05/16 09:47:17] user-mgr.c(773): Ldap init and bind failed using ‘cn=XXX,dc=XXX,dc=XXX': ‘XXXXXXX' on server 'ldaps://10.XX.XX.XX/'.
If you are using pro edition, you can check the LDAP configuration by running a script as described in http://manual.seafile.com/deploy_pro/using_ldap_pro.html (search Testing your LDAP Configuration).
If the script can correctly talk to ldap server, it is most likely caused by incompatible of bundled LDAP libraries. You can follow http://manual.seafile.com/deploy/using_ldap.html (the end of document) to remove the bundled LDAP libraries.
How to restrict Seafile access to certain accounts in AD
You can use FILTER field in LDAP configuration in
ccnet.conf. For example, the following filter restricts the access to Seafile to members of a group.
FILTER = memberOf=cn=group,cn=users,DC=x
AD also supports subgroups. The following filter restricts the access to Seafile to membersand subgroups of a group.
FILTER = memberOf:1.2.840.113518.104.22.1681:=cn=group,cn=users,DC=x
For more information on the Filter syntax, see http://msdn.microsoft.com/en-us/library/aa746475%28VS.85%29.aspx
After upgrading Web UI is broken because CSS files can't be loaded
Please remove the cache and try again,
rm -rf /tmp/seahub_cache/*. If you configured memecached, restart memcached, then restart Seahub.
If the problem is not fixed, check whether seafile-server-latest point to the correct folder. Then check whether
seafile-server-latest/seahub/media/CACHE is correctly being generated (it should contain the auto-generated CSS file(s)).
Avatar pictures vanished after upgrading the server, what can I do?
You need to check whether the "avatars" symbolic link under seahub/media/ is linking to ../../../seahub-data/avatars. If not, you need to correct the link according to the "minor upgrade" section in Upgrading-Seafile-Server.
If your avatars link is correct, and avatars are still broken, you may need to refresh Seahub cache using
rm -rf /tmp/seahub_cache/*or by restarting memcached if being used.
System can't start
Seahub can't be started after power failure
Please try clean the cache directory
Please also check the permission of
seahub.log. If Seahub can't write to these files, it will fail to start.
Seafevents can't be started
Office files online preview can't work. There is no logs in seafevents.log. From
controller.log, the seafevent process is being started again and again.
Please check the permission of
seafevent.log. If seafevent can't write to these files, it will fail to start.
You can also try start seafevents manually following: https://download.seafile.com/f/423d5d6301/
How to migrate libraries and groups from one account to another?
Since version 4.4.2, system admins can migrate libraries and groups from one account to another existing account using RESTful web api.
Seafile GC shows errors, FSCK can’t fix them
GC scans the history. But FSCK only scans the current version. You can ignore the error. It is a minor issue.
Ceph and S3
Seafile server can't started when using Ceph
Seafile server can't started when using Ceph as storage backend. seafile.log is empty. controller.log shows:
[10/20/16 12:39:29] seafile-controller.c(568): pid file /opt/seafile/pids/seaf-server.pid does not exist [10/20/16 12:39:29] seafile-controller.c(588): seaf-server need restart... [10/20/16 12:39:29] seafile-controller.c(198): starting seaf-server ...
This is most likely caused by Ceph library incompatible.
You can install
librados provided by the Linux distribution, then remove bundled libraries:
cd seafile-server-latest/seafile/lib rm librados.so.2 libstdc++.so.6 libnspr4.so
You can also run
seaf-fsck.sh. It will print detailed message if Ceph library incompatible.
GC error when removing blocks in Ceph
We just did a GC run which came up with errors when deleting blocks. This seems to happen with all blocks/libraries. Below is an example for a single library.
Starting seafserv-gc, please wait ... [08/29/16 09:15:41] gc-core.c(768): Database is MySQL/Postgre, use online GC. [08/29/16 09:15:41] gc-core.c(792): Using up to 10 threads to run GC. [08/29/16 09:15:41] gc-core.c(738): GC version 1 repo Documents(135ca71c-da2b-4b07-86e3-c7a1d46b9b22) [08/29/16 09:16:04] gc-core.c(510): GC started for repo 135ca71c. Total block number is 294. [08/29/16 09:16:04] gc-core.c(68): GC index size is 1024 Byte for repo 135ca71c. [08/29/16 09:16:04] gc-core.c(269): Populating index for repo 135ca71c. [08/29/16 09:16:04] gc-core.c(334): Traversed 33 commits, 402 blocks for repo 135ca71c. [08/29/16 09:16:04] gc-core.c(559): Scanning and deleting unused blocks for repo 135ca71c. [08/29/16 09:16:04] ../../common/block-backend-ceph.c(463): [block bend] Failed to remove block 79fc986a: No such file or directory. [08/29/16 09:16:04] ../../common/block-backend-ceph.c(463): [block bend] Failed to remove block ae2678f8: No such file or directory. [08/29/16 09:16:04] ../../common/block-backend-ceph.c(463): [block bend] Failed to remove block 9fe1ca0b: No such file or directory. [08/29/16 09:16:04] ../../common/block-backend-ceph.c(463): [block bend] Failed to remove block 4cad277e: No such file or directory. [08/29/16 09:16:04] ../../common/block-backend-ceph.c(463): [block bend] Failed to remove block e9c94b16: No such file or directory. [08/29/16 09:16:04] gc-core.c(577): GC finished for repo 135ca71c. 294 blocks total, about 402 reachable blocks, 5 blocks are removed. [08/29/16 09:16:04] gc-core.c(839): === GC is finished === seafserv-gc run done
Your "issue" looks similar to the one discussed here: http://lists.ceph.com/pipermail/ceph-users-ceph.com/2015-November/005837.html
That should be related to the behavior of cache tier in Ceph. You could try to use "rados rm" command to remove that object. If it returns the same error (no such file or directory), it should be the same issue. You should try to copy that object out before removing it, in case you still need it later.
Page layout broken because seahub/media/CACHE is created only on first node
COMPRESS_CACHE_BACKEND = 'django.core.cache.backends.locmem.LocMemCache'
seahub_settings.py as documented at http://manual.seafile.com/deploy_pro/deploy_in_a_cluster.html
This is going to tell every node to generate the CSS CACHE in its local folder.