Re: Blew away .../var/varnish/HOSTNAME/ -- broke varnishstat, how to recover?
On Mon, Jun 2, 2008 at 7:57 AM, Chris Shenton <[EMAIL PROTECTED]> wrote: > We have to fill out pounds of paperwork in order to take any outage on > a public server, no matter how small. Is there a way to restart > Varnish without any downtime -- to continue accepting but holding > connections until restarted, rather like Apache's "apachectl graceful" > does? Other ideas? Can you avoid the problem by putting your Varnish servers behind a load balancer? That way, you can preemptively disable the server from taking traffic on the LB side prior to restarting Varnish, thereby eliminating any perceivable customer effect. Also, be careful about using "apachectl graceful" (especially combined with log rotation), as connections that are currently idle but which may never receive traffic again will not be terminated. I consider it too leaky to use. Best regards, --Michael ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Varnish/Swap usage and network timeouts
Hello,
I have 2 machines running varnish which handles around 1000 requests per
second each and sometimes the machines will just stop responding to any
network communication. A coworker used wireshark to check the connection
and saw a whole lot of tcp retransmissions. Normally I would suspect
some sort of network buffer issue but there is nothing in the logs at
all. This is on FreeBSD 6.3 with a default 64 bit kernel with bge
network cards. They are both running varnish-trunk revision 2635.
Attached is the VCL file that is currently running. Varnish is started
with the following options:
file,/vol/data1/varnish.cache,90% -t 259200 -h classic,59 -p
lru_interval=3600
The only suspicious things I've found are: Varnish says it is using 372G
of virtual memory when I told it to use a 58gb file and there is only
2gb of physical memory in the machine and there is pretty constant
swapping going on, mostly under 500K but it seems to be pretty constant
no matter the number of connections and such. The amount of swap space
in the machine is 4gb and things are using 1.1gb.
Threads don't seem to be a problem, as the highest I've seen it is at
500 and that was after an extended period of time around 250 or so. So I
don't think I'm running into thread issues
So I guess I'm asking if anyone has any other suggestions other then
reinstalling to FreeBSD 7.0 which seems to be the best solution I can
see. If that is the best solution, what is the optimal FreeBSD
configuration for Varnish? Lots of swap space? No swap space? Currently
the mmap-ed file is on an SSD so one alternative I thought of was to
make the entire SSD part of swap and use the malloc option and get rid
of any regular disk basked swap.
Thanks for any help,
Erik
backend default {
.host = "xzy.com";
.port = "80";
}
sub vcl_recv {
if (req.url ~ "\.js|css$") {
if (req.http.Accept-Encoding ~ "gzip") {
set req.http.Accept-Encoding = "gzip";
} elsif (req.http.Accept-Encoding ~ "deflate") {
set req.http.Accept-Encoding = "deflate";
} else {
unset req.http.accept-encoding;
}
} else {
unset req.http.accept-encoding;
}
unset req.http.Cookie;
set req.grace = 10m;
}
sub vcl_fetch {
set obj.http.P3P = "CP='NOI CURa ADMa DEVa TAIa OUR BUS IND UNI COM
NAV INT', policyref='http://www.okcupid.com/w3c/p3p.xml'";
set obj.grace = 10m;
if (obj.ttl < 259200s) {
set obj.ttl = 259200s;
}
}
___
varnish-misc mailing list
varnish-misc@projects.linpro.no
http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: Blew away .../var/varnish/HOSTNAME/ -- broke varnishstat, how to recover?
On Jun 2, 2008, at 11:19 AM, Poul-Henning Kamp wrote: > > Yes, restarting varnish (completely: ie both manager and child > process) > should recreate it with no side effects. > > Varnish doesn't spend long time on startup, so I would just get it > done and over with. Just tried it on a QA server and it restarts in about 0.25 seconds, and re-creates the directory with the _.vsl file. Have to see if my overlords are OK with that. Thanks. > > > (In the meantime, if you have enabled the telnet option, you can still > see the stats through the CLI interface.) I do have that and can ask it for stats. Thanks for the reminder. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: Blew away .../var/varnish/HOSTNAME/ -- broke varnishstat, how to recover?
In message <[EMAIL PROTECTED]>, Chris Shenton write s: >What's the recommended way to re-create this directory? I'm assuming >restarting Varnish will re-create it but want to be sure. Yes, restarting varnish (completely: ie both manager and child process) should recreate it with no side effects. >We have to fill out pounds of paperwork in order to take any outage on >a public server, no matter how small. Is there a way to restart >Varnish without any downtime -- to continue accepting but holding >connections until restarted, rather like Apache's "apachectl graceful" >does? Other ideas? Varnish doesn't spend long time on startup, so I would just get it done and over with. You are in no rush however, so you can schedule it for whenever you want. (In the meantime, if you have enabled the telnet option, you can still see the stats through the CLI interface.) -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Blew away .../var/varnish/HOSTNAME/ -- broke varnishstat, how to recover?
We're using varnish for a public site, works beautifully and handled getting slashdotted gracefully -- many thanks! We built it with "zc.buildout" which creates .../parts/varnish/install/ var/varnish/FQDN/... and others. In a recent update, I accidentally blew away the .../var/varnish/FQDN directory which contained a few files that seem necessary for "varnishstat" and perhaps others to work. bin.XXkP5ZYv _.c _.vsl $ varnishstat Cannot open /usr/local/mastersite_buildout/parts/varnish/install/var/ varnish/FQDN/_.vsl: No such file or directory Varnish continues to serve well, so I haven't restarted. What's the recommended way to re-create this directory? I'm assuming restarting Varnish will re-create it but want to be sure. We have to fill out pounds of paperwork in order to take any outage on a public server, no matter how small. Is there a way to restart Varnish without any downtime -- to continue accepting but holding connections until restarted, rather like Apache's "apachectl graceful" does? Other ideas? Thanks. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: Conditional caching question
In message <[EMAIL PROTECTED]>, David Pratt writes: >Hi. In most cases, I want a request to be passed to a backend where it >will be handled by server. If frequency is high, however; I want to add >the object to varnish cache and have varnish handle it. I am not worried >about a mechanism to keeping track of frequency of requests. Question is >what is available to me to add an object/path to the varnish cache if it >was originally passed? That's the problem, we don't have any place to hold statistics for objects that get "pass" treatment, so we would have no way of knowing that a particular object was suitable for caching. You could write a program that extracts that information from the varnishlog on the fly (sort of like varnishtop), but then we need a mechanism to get from there to tell varnishd that this or that object should be cached from now on. I wouldn't say that your way of using varnish is backwards relative to the design objectives, but you do come close, since we assumed caching by default, and pass as exception, rather than the other way around. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
