Minor changes in this one. A rename worked fine. While at it, we should fix the nit described in bug 410645.
I'll add a patch to support upgrades too (bug 410647).
I'll bump as soon as I get an updated init script from you, Johan.
Thanks. Sorry for the delay; I've been looking into the init script and see some discrepancies: 1. pid should go into /var/run/pgbouncer , but the ebuild doesn't create that directory any more (or point to it through init script) 2. upstream config now has saner default values (fix sed in ebuild) 3. the /run vs /var/run check needs to be improved; I use mdev over udev and have an empty /run which makes the init script bail 4. add "upgrade" to init script Status: 1,2 is fixed. 3 I don't know the best solution for. 4 I would just like to skip since I've had more issues with this in production. If the user really really wants it, he/she could call it manually. Thoughts?
Does anyone have any suggestions on how to proceed with /run,/var/run situation? I use mdev/devtmpfs over udev which probably creates my /run , but its empty. Is there a gentoo pattern for handling this?
(In reply to comment #4) > Does anyone have any suggestions on how to proceed with /run,/var/run > situation? I use mdev/devtmpfs over udev which probably creates my /run , > but its empty. Is there a gentoo pattern for handling this? Yes, see what I've done in dev-db/pgpool2.
Created attachment 328408 [details] pgbouncer-1.5.3.ebuild Lets update the current ebuild to 1.5.3 to address the security bug 440940. I will (finally) get back to you with an -r1 that addresses all known issues later this week.