Question & Answer
Question
Seeing numerous EDC6257C THERE IS A SOCKET PROBLEM, SYSLOGD MAY NEED TO BE RESTARTED messages in the system log. At the same time, numerous server applications (such as INETD, FTPD, IKED, ...) are experiencing responsiveness delays and failures.
Answer
The action to fix the immediate problem is to recycle the SYSLOGD address space (typically a started task).
The typical cause of this condition is having a system that is still using the older method of archiving SYSLOGD files that uses a process (CRON task or other automation facility) to send a SIGHUP signal to the SYSLOGD process (typically at midnight). If this signal arrives in the middle of other scheduled activity within SYSLOGD, it can cause a deadlock by terminating a thread that other processes are waiting to complete. The times at which this scheduled activity occurs are controlled by the ArchiveTimeOfDay and ArchiveCheckInterval statements in the /etc/syslog.conf file (whether or not the built-in archival function is being used).
To avoid this problem, either:
Convert to using the built-in archival processing , and remove the processing that sends the SIGHUP (and performs any related file functions).
Change the ArchiveTimeOfDay and ArchiveCheckInterval values so that the resultant schedule avoids the time when the SIGHUP is performed.
Product Synonym
ZOSCS COMMSERVER
Was this topic helpful?
Document Information
Modified date:
20 February 2017
UID
dwa1251764