Bug #2592
closedFCGI backend not reactivated if time change occurs during 1s delay
Description
When an fcgi backend dies, lighttpd disables it for 1s before trying again.
If, during this 1s delay, the system time is changed backwards for e.g. 2 hours, lighttpd will wait for 2 hours before re-enabling the backend.
2014-08-28 18:51:33: (mod_fastcgi.c.1733) connect failed: Connection refused on unix:<socket-file>
2014-08-28 18:51:33: (mod_fastcgi.c.2999) backend died; we'll disable it for 1 seconds and send the request to another backend instead: reconnects: 0 load: 1
<at this point, the system time was changed backwards by 2 hours>
2014-08-28 16:51:49: (mod_fastcgi.c.3565) all handlers for <X> on <Y> are down.
2014-08-28 18:51:35: (mod_fastcgi.c.2826) fcgi-server re-enabled: unix:<socket-file>
lighttpd/1.4.34
Updated by stbuehler over 10 years ago
- Status changed from New to Wontfix
- Target version deleted (
1.4.x)
This is a problem with nearly all server software, and hard to work around. Just run ntp as a server instead of ntpdate, and it will adjust the time slowly instead of forcing time jumps.
This is unlikely to get fixed with the current timer management.
Updated by gstrauss over 1 year ago
- Status changed from Wontfix to Fixed
- ASK QUESTIONS IN Forums set to No
Fixed in lighttpd 1.4.60. Since lighttpd 1.4.60, lighttpd, where practical, uses monotonic time obtained from the system.
Also available in: Atom