lighttpd spontaneously dies

Question

I've been using lighttpd pretty successfully on a VPS for several years, but I've hit a snag since upgrading from Ubuntu 14.04 to 16.04.1 on September 30th. (My lighttpd is now 1.4.35.) Twice, upon trying to visit my website, I've found lighttpd not running even though the server itself is still up. lighttpd runs fine once I start it up again. For the second incident, all I have in my lighttpd error logs is:

2016-10-09 14:10:57: (connections.c.1702) SSL: 1 -1 error:140E0197:SSL routines:SSL_shutdown:shutdown while in init 
2016-10-10 02:12:55: (connections.c.305) SSL: 1 error:1408A10B:SSL routines:ssl3_get_client_hello:wrong version number 
2016-10-10 02:14:01: (connections.c.305) SSL: 1 error:1408A0C1:SSL routines:ssl3_get_client_hello:no shared cipher 
2016-10-10 06:48:22: (connections.c.305) SSL: 1 error:140943E8:SSL routines:ssl3_read_bytes:reason(1000) 
2016-10-10 06:48:22: (connections.c.1702) SSL: 1 -1 error:140E0197:SSL routines:SSL_shutdown:shutdown while in init 
2016-10-10 07:01:39: (connections.c.1702) SSL: 1 -1 error:140E0197:SSL routines:SSL_shutdown:shutdown while in init 
2016-10-10 07:35:02: (server.c.1558) server stopped by UID = 0 PID = 7365 
2016-10-10 07:35:02: (log.c.164) server started 
2016-10-10 07:35:02: (server.c.1558) server stopped by UID = 0 PID = 1

breakage.log is empty.

Any hints on how I can figure out what's killing the process? PID 1 is systemd, but PID 7365 is dead by now, and I don't know what it was.

Anything in the system log file (/var/log/syslog)? Could the system be running out of memory? — Paul Haldane, Oct 12 '16 at 06:33
@PaulHaldane /var/log/syslog is empty. `top` shows plenty of free memory right now, but I guess that doesn't rule out anything. — Kodiologist, Oct 12 '16 at 14:21

score 0 · Accepted Answer · answered Oct 17 '16 at 18:07

The spontaneous death happened again yesterday, at the same time of day (but on a different day of the week):

2016-10-16 07:35:02: (server.c.1558) server stopped by UID = 0 PID = 2169
2016-10-16 07:35:03: (log.c.164) server started 
2016-10-16 07:35:03: (server.c.1558) server stopped by UID = 0 PID = 1

I think the problem was related to my custom /etc/logrotate.d/lighttpd, which used the old command invoke-rc.d lighttpd reload instead of the new command invoke-rc.d lighttpd reopen-logs. I've changed to reopen-logs, and hopefully, the problem won't reoccur.

lighttpd spontaneously dies

1 Answers1