The log files of seafile-server are very hard to understand for a system administrator. First, a lot of log output is not interesting. Examples:
ccnet.log:
[04/18/15 18:28:46] ../common/session.c(375): Accepted a local client
[04/18/15 18:28:47] ../common/session.c(375): Accepted a local client
[04/18/15 18:29:05] ../common/peer.c(943): Local peer down
[04/18/15 18:29:05] ../common/processor.c(218): [Proc] Shutdown processor threaded-rpcserver-proc(-1001) for bad update: 515 peer down
[04/18/15 18:29:06] ../common/session.c(375): Accepted a local client
[04/18/15 18:29:07] ../common/peer.c(943): Local peer down
[04/07/15 11:36:11] ../common/mq-mgr.c(60): [mq client] mq cilent is started
[04/07/15 11:36:11] ../common/mq-mgr.c(106): [mq mgr] publish to hearbeat mq: seaf_server.heartbeat
[04/07/15 11:36:11] listen-mgr.c(115): listen on port 12001 for block tranfer
These messages may be important for developers, but as a system administrator I wouldn't know what to make of them. At the same time many important things are happening that leave no trace in the log files at all, like:
A user is created/deleted/logs in/out
A repository is created/deleted
blocks/commits/fs are created/removed
clients connected, with remote IP address
database handling
startup/shutdown
helpful error messages in case of errors
(many more things)
So I'd like to suggest:
to add loglevels: i.e. trace, debug, info, warn, error
to add a config switch to change the log level
to review the log messages and to sort them into above categories
to add more log messages. To give adminsitrators a feeling of what seafile is currently doing. And get clues if something goes wrong.
The log files of seafile-server are very hard to understand for a system administrator. First, a lot of log output is not interesting. Examples:
ccnet.log:
[04/18/15 18:28:46] ../common/session.c(375): Accepted a local client [04/18/15 18:28:47] ../common/session.c(375): Accepted a local client [04/18/15 18:29:05] ../common/peer.c(943): Local peer down [04/18/15 18:29:05] ../common/processor.c(218): [Proc] Shutdown processor threaded-rpcserver-proc(-1001) for bad update: 515 peer down [04/18/15 18:29:06] ../common/session.c(375): Accepted a local client [04/18/15 18:29:07] ../common/peer.c(943): Local peer down
controller.log:
[04/10/15 11:18:04] seafile-controller.c(88): spawned seaf-server, pid 2953 [04/10/15 11:18:04] seafile-controller.c(529): seafdav not enabled. [04/11/15 12:44:56] seafile-controller.c(154): starting ccnet-server ... [04/11/15 12:44:56] seafile-controller.c(73): spawn_process: ccnet-server -c /usb-hdd/seafile/ccnet -f /usb-hdd/seafile/logs/ccnet.log -d -P /usb-hdd/seafile/pids/ccnet.pid [04/11/15 12:44:56] seafile-controller.c(88): spawned ccnet-server, pid 2960 [04/11/15 12:44:57] seafile-controller.c(556): ccnet daemon connected. [04/11/15 12:44:57] seafile-controller.c(185): starting seaf-server ...
seafile.log:
[04/07/15 11:36:11] ../common/mq-mgr.c(60): [mq client] mq cilent is started [04/07/15 11:36:11] ../common/mq-mgr.c(106): [mq mgr] publish to hearbeat mq: seaf_server.heartbeat [04/07/15 11:36:11] listen-mgr.c(115): listen on port 12001 for block tranfer
These messages may be important for developers, but as a system administrator I wouldn't know what to make of them. At the same time many important things are happening that leave no trace in the log files at all, like:
So I'd like to suggest: