![]() |
lvs-users
|
| To: | Roberto Nibali <ratz@xxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] reinstate state transition timer setting in proc-fs (resend v2) |
| Cc: | "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx> |
| Cc: | Horms <horms@xxxxxxxxxxxx> |
| From: | Julian Anastasov <ja@xxxxxx> |
| Date: | Wed, 7 Dec 2005 00:56:23 +0200 (EET) |
Hello,
On Tue, 6 Dec 2005, Roberto Nibali wrote:
> So, reading the code a bit more I realise that a basic infrastructure
> has been prepared by extending all ip_vs_* structs with int* timeout
> hooks. How do you want to solve passing such variable information via
> ipvsadm? I would really like to read your proposal on this granular
> timeout framework.
Here is what i found in sent-mail archives:
http://www.ssi.bg/~ja/tmp/tt.txt
Currently, all protocols provide names for their state, so
little support for per-proto timeouts is required in kernel. More
code is needed for per-service timeouts or even per-app (it has
.timeout_table). But it needs thinking and time.
Regards
--
Julian Anastasov <ja@xxxxxx>
|
| Previous by Date: | Re: ipvsadm --set... what does it modify?, Guy Waugh |
|---|---|
| Next by Date: | outbound nat problem, Rob Ruth |
| Previous by Thread: | Re: [PATCH] reinstate state transition timer setting in proc-fs(resend v2), Roberto Nibali |
| Next by Thread: | Re: [PATCH 2.4][RFC] hprio scheduler for atomic server pool switching, Horms |
| Indexes: | [Date] [Thread] [Top] [All Lists] |