Thanks for reply.
Earlier we had the model you described above. tail -n0 -F /var/log/httpd/error_log /var/log/httpd/ssl_error_log | java program We experienced that there is some message loss, meaning java program was not able to get few logs very irregularly. Though the number was very less we needed to fix it. So we reduced one hop by getting rid if the tail. And as I said, it was working with the setting mentioned in original question. It used to spin off new program after terminating previous one. But it suddenly stopped doing that. On Thu, Mar 23, 2017 at 5:44 AM, Houser, Rick <rick.hou...@jackson.com> wrote: > Stick with rotatelogs and adjust your program (or provide a separate > helper) to read the logs from disk as a traditional log parser? That way, > you have standard logging behavior, your customizations aren’t altering > httpd behavior, and you could even drop your utility onto an existing > Apache server with no config changes. > > > > > > Rick Houser > > Web Administration > > > > *From:* Milind Vaidya [mailto:kava...@gmail.com] > *Sent:* Wednesday, March 22, 2017 17:31 > *To:* users@httpd.apache.org > *Subject:* [users@httpd] external log reader not getting restarted every > hour > > > > > > I have following setting in my *httpd.conf* > > *ErrorL**og* *"| /usr/bin/tee -a /var/log/httpd/error_log | java program > reading msgs from stdinput"* > > The error_log file gets rotated using logrotate every hour. > > The java program has following statement > > *while** ((msg = **this**.stdinReader.readLine()) != **null**) {* > > *}* > > This essentially reads the lines till EOF is encountered. After this I > close all the resources and the main method ends. > > Earlier I was NOT ending the program when EOF was encountered. So apache > used to start new process every hour keeping old one intact. But this > stopped happening when the logic to terminate program was added when EOF is > encountered. > > > > But now this does not seem to work with everything else remaining same. > Now the same process keeps on running for days together which is little > worrisome as does not look like the normal behaviour. > > > > What are my debugging options here. >
