user login error: Error reading socket, ret=0, errno=0.(778)

[hubnb]

today,our user login QAD system error:" Error reading socket, ret=0, errno=0.(778)".
I just check mfgprod.lg ,get error message :
20:37:22 SRV 9: errno=29 reading promsgs file, it may have been deleted.
20:37:22 SRV 9: Unable to format message number 74


and mfgprod.lg 's size is 4.8G .

my progress version is 91e.

I have no idea with the problem , please help me.

 

[Cringer]

Has anything changed in the last 24 hours?

 

[RealHeavyDude]

The knowledge base is pretty elaborate on the error 778:

http://knowledgebase.progress.com/articles/Article/P103854

.

The error with the promsgs just tells that the promsgs file can't be read. Either because it does not exist or you don't have read permissions. If not specified otherwise the promsgs file resides in the installation directory. This file holds the message text for all Progress error numbers. During installation it is copied from one of the variants in the prolang sub directories according to what you've chose as language. You can specify to use a different promsgs file by setting the environment variable PROMSGS. Usually this error indicates that the environment is not setup correctly. Either because the DLC environment is missing or does not point to the installation directory and your PATH does not point into the Progress installation either.

Since it did work yesterday I would speculate that somebody fiddled either with the environment variables or with the installation directory.

Heavy Regards, RealHeavyDude.

 

[TomBascom]

Errno = 29 is "invalid seek".

Virus scans can sometimes make PROMSGS unreadable as well.

A 4GB .lg file is a bad thing. Truncate it.

And, of course, 9.1E is ancient, obsolete and effectively unsupported. You should upgrade.

 

[CarlosO]

We have the "Error reading socket 778 ..." since August 2011 with 9.1D. It's really frustating because the KB talks about a TCP/IP communication problem and there is no workaround. We have it since we introduced Windows 7 machines, before that we have no problem in almost 10 years with our "ancient, obsolete and unsopported" but for us venerable 9.1D. It's curious, checking the KB we found the same problem is in version 10.x and maybe 11.x so it appears upgrading it's not (necessarily) the solution.

 

[Rob Fitzpatrick]

We have the "Error reading socket 778 ..." since August 2011 with 9.1D. It's really frustating because the KB talks about a TCP/IP communication problem and there is no workaround. We have it since we introduced Windows 7 machines, before that we have no problem in almost 10 years with our "ancient, obsolete and unsopported" but for us venerable 9.1D. It's curious, checking the KB we found the same problem is in version 10.x and maybe 11.x so it appears upgrading it's not (necessarily) the solution.

What platform are your databases running on? Are there any errors on that side, e.g. protrace files? Do you have network technicians who can capture and analyze traffic?

 

[CarlosO]

It's a Windows platform. I've been analyzing the logs in the progress and windows side but don't see anything strange. I doubt it's a hardware network problem.
The error message appears in some specific programs. I identified the programs and lines of code (hard work, because the error only appears 3 or 4 times a month and sometimes last 5 minutes, sometimes last 5 hours) which provoked the error and replaced them.
Guess what? No more errors. Really strange, the same program running without problems for years and after Windows 7 appears the "Error reading socket ...".
In the first case was a 'If ... then ... next' after a 'For each ...', i included the question of the 'if' in the 'For each' and no more error.
The second case was a 'Find ...' using a field not indexed, i forced the use of a indexed field and no more error.
I still have five programs to trace so i'm waiting for next time the error appears.