Problem restarting httpd after certificate request

[andreagobetti]

When the script tries to restart httpd after a certificate request I have the following error:

Starting httpd: (98)Address already in use: make_sock: could not bind to address xxx.xxx.xxx.xxx:8080
no listening sockets available, shutting down
Unable to open logs

To resolve the problem I used:

lsof -i :8080

to see who was binding port 8080, and then i terminated manually the httpd processes that were listening on port 8080 using

kill -SIGTERM <pid>

I'm not sure it is strictly related with this project, but if it's not this can be helpful for other users experiencing similar problems

My server is running: CentOS release 6.7 Vesta 0.9.8 Apache 2.2.15 nginx 1.10.0

[jpitoniak]

It kind of sounds like you have something going on that's causing Apache to hang when it tries to shutdown. While I haven't seen this particular issue on any Vesta server, I have seen Apache hiccup on restarts in other environments before.

Two things you might try:

• Try doing a "service httpd stop" after you run the letsencrypt-vesta script and get an error. That should cleanly stop all Apache processes. Give it a few seconds and then do a "service httpd start" to restart it.
• You might also try changing the "service httpd restart" to "service httpd reload" in the letsencrypt-vesta script. Technically, a full Apache restart shouldn't be necessary to load the new cert; a reload of the configs should work in most cases.

[andreagobetti]

Hi, thanks for your reply!

• I noticed that service httpd restart has this kind of problem only after the execution of letsenrcypt-vesta
• After letsencrypt-vesta execution service httpd stop fails, so unfortunately this doesn't help
• Changing service httpd restart in service httpd reload solved the problem!

[exaexa]

+1

Please change to reload, it solves the problem for me too. IIRC there's no need to restart the server when only changing the SSL params.