Error log full of "Socket not active"

[johnnyt]

Last night I was noticing performance problems and checked the error log, a ~4MB file full of messages like this:

 

Sat 2013/04/06 08:34:28 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:28 PM System -170001 [HTTP] 192.168.0.55:1465->80

Sat 2013/04/06 08:34:29 PM System -170001 0: GET-->/rest/vars/set/2/52/348

Sat 2013/04/06 08:34:29 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:32 PM System -170001 [HTTP] 192.168.0.55:1466->80

Sat 2013/04/06 08:34:32 PM System -170001 0: GET-->/rest/vars/set/2/58/1850

Sat 2013/04/06 08:34:32 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:34 PM System -170001 [HTTP] 192.168.0.55:1467->80

Sat 2013/04/06 08:34:34 PM System -170001 0: GET-->/rest/vars/set/2/59/2068

Sat 2013/04/06 08:34:34 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:35 PM System -170001 [HTTP] 192.168.0.55:1468->80

Sat 2013/04/06 08:34:35 PM System -170001 0: GET-->/rest/vars/set/2/61/2125

Sat 2013/04/06 08:34:35 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:37 PM System -170001 [HTTP] 192.168.0.55:1469->80

Sat 2013/04/06 08:34:37 PM System -170001 0: GET-->/rest/vars/set/2/51/181

Sat 2013/04/06 08:34:37 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:38 PM System -170001 [HTTP] 192.168.0.55:1470->80

Sat 2013/04/06 08:34:38 PM System -170001 0: GET-->/rest/vars/set/2/50/2468

Sat 2013/04/06 08:34:38 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:40 PM System -170001 [HTTP] 192.168.0.55:1471->80

Sat 2013/04/06 08:34:40 PM System -170001 0: GET-->/rest/vars/set/2/62/968

Sat 2013/04/06 08:34:40 PM System -170001 [HTTP] Socket not active#2 27

Sat 2013/04/06 08:34:41 PM System -170001 [HTTP] 192.168.0.5:11238->80

Sat 2013/04/06 08:34:41 PM System -170001 [HTTP] 192.168.0.5:11241->80

Sat 2013/04/06 08:34:42 PM System -170001 [HTTP] 192.168.0.55:1472->80

Sat 2013/04/06 08:34:42 PM System -170001 0: GET-->/rest/vars/set/2/60/2093

Sat 2013/04/06 08:34:50 PM System -170001 [HTTP] 192.168.0.5:11242->80

Sat 2013/04/06 08:34:53 PM System -170001 [HTTP] Socket not active#2 27

 

...going back to Friday 2 AM Friday (about 42 hours worth)

 

FWIW, at 192.168.0.55 is an Autelis RS232-to-ISY bridge reading data from my 1-wire temp/humidity sensor solution, and at 192.168.0.5 is the Admin GUI, which I leave running 24/7

 

Restarting the GUI fixed the problem. Does that mean it's what lived at socket #2 27? Are the entries happening at the same time related or unrelated and just coincidental? How do I find out for sure what lives at socket #2 27? Why doesn't ISY (or the subscriber to ISY?) close the socket after a period of time?

 

Also, I understand from a previous post that having the event viewer at level 3 can fill the error log with these messages and I did have the event viewer at level 3 for a few hours yesterday afternoon, but before that (going back to 2AM Friday) it was at the default level 1. Does that mean the messages are warnings rather merely informational, as I understand they usually are?

 

Finally, at what point (if any) does the error log drop the oldest entry to make room for the next one, e.g. is it by size or number of lines?

 

Any info would be appreciated.

[Michel Kohanim]

Hi johnnyt,

 

They are informational and diagnostic messages related to Event Viewer > 1. That's why closing the Admin Console fixes it.

 

If you telnet to ISY, login, and type '?', you get the menu. CL allows you to configure the MAX log size the default for which is 3 MB right now. Please note that 3 MB is the size of the file in ISY and NOT what you save in your computer (which is larger) since Admin Console goes through all entries and makes a few things more human readable such as date/time.

 

With kind regards,

Michel

[johnnyt]

Thanks for the info.

 

I'm a little confused because I know I did not have the event viewer on level 3 for the entire period from Friday 2 AM until I restarted the GUI at 8:30 last night. I specifically had to change the level to 3 yesterday (Sat) afternoon to verify communications in my kitchen area so I know it was at level 1 for most of the time the log was filling up. Not to say I hadn't changed it to 3 at some point in the past but it was at level 1 between Fri 2 AM and yesterday afternoon. Maybe I'll do some testing to see if once the event viewer gets set to 3 it remains there for logging purposes regardless of what I set it back to subsequently.

 

What about sockets not being active for extended periods of time? I looked back and 27 was not the only socket not active; there were several others in the log file. I didn't review them all but a few I followed from start to finish were "not active" for 4-5 hours. Much of the time the log file was getting hammered with entries 3-6 times per second with only 1 or 2 second pauses in between.

 

Is it ISY that doesn't close the socket or a subscriber to ISY? How do I find out which subscriber(s) is/are involved? This amount of logging of non active sockets has me worried there's a problem with my setup and I'm wondering if the problem is at the root of the two crashes I encountered less than a month ago (see viewtopic.php?f=25&t=11001), which remain a mystery and included both a lost error log file and a full (4.7MB) error log...

[Michel Kohanim]

johnnyt,

 

I guarantee you that those entries with IP address are because the debug level is above 2.

 

Socket not active is just a warning. It basically means socket was not closed properly and ISY keeps it in a list and will eventually close and release resources. I would not worry about them UNLESS it's the SAME socket information repeated for more than 10 minutes.

 

From my perspective, I really do not see any problems in the logs. They look pretty normal to me. Obviously, you are going to have a very little performance hit if you are having Event Viewer on level 3 but I really do not think these are the source of performance degradation of your system.

 

With kind regards,

Michel

[johnnyt]

Socket not active is just a warning. It basically means socket was not closed properly and ISY keeps it in a list and will eventually close and release resources. I would not worry about them UNLESS it's the SAME socket information repeated for more than 10 minutes.

 

I've got HOURS of log entries about socket 27, 32, 33 and others but can't tell if all entries are about the same socket 27, 32 and 33 throughout, or if a new socket just ends up with an old number. Should I send you the ~4MB file for you to see? Alternatively, can you add an informational message that shows when a socket is dropped in the next beta?

[Michel Kohanim]

Hi johnnyt,

 

It's really normal for sockets not to close and, as such, I really wouldn't worry about them. If it makes you feel better, please clear your log, make sure you do not have event viewer running, and then send me your log after two days to support@universal-devices.com.

 

With kind regards,

Michel

[johnnyt]

Hi johnnyt,

 

It's really normal for sockets not to close and, as such, I really wouldn't worry about them. If it makes you feel better, please clear your log, make sure you do not have event viewer running, and then send me your log after two days to support@universal-devices.com.

 

With kind regards,

Michel

What would make me FEEL better is knowing why ISY crashed twice a month ago, once apparently deleting or corrupting the error log and once after filling one up within a few days. I thought maybe I was getting closer but I guess I'm not.

[Michel Kohanim]

Hi johnnyt,

 

You and I both ... I very much wish to know what caused the crash.

 

With kind regards,

Michel