Date: Fri, 10 Apr 2015 11:01:44 +0200 (CEST) From: Raimund Sacherer <rs@logitravel.com> To: opendaddy@hushmail.com Cc: freebsd-questions@freebsd.org Subject: Re: Initial request to server extremely slow after longer periods of inactivity Message-ID: <369199529.123430074.1428656504549.JavaMail.zimbra@logitravel.com> In-Reply-To: <20150409185801.1B1E9401E3@smtp.hushmail.com> References: <20150409034121.7D7C720395@smtp.hushmail.com> <CADqw_gL6b4fYyda9GzKt4BGv=d=sASttf4bp2_W%2B6JnuKQUCjg@mail.gmail.com> <20150409113928.EE31F401E4@smtp.hushmail.com> <CADqw_gK8-bxeNTYw4ATQnSM-w0iNBVWma83hMScdbsCwzEGxEw@mail.gmail.com> <20150409185801.1B1E9401E3@smtp.hushmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
----- Original Message -----=20 > From: opendaddy@hushmail.com > To: "Michael Schuster" <michaelsprivate@gmail.com>, terje@elde.net, > freebsd-questions@freebsd.org > Sent: Thursday, April 9, 2015 8:58:00 PM > Subject: Re: Initial request to server extremely slow after longer period= s of > inactivity > It's probably not my DNS. From the NSD mailinglist: > > On 9. april 2015 at 6:49 PM, "Ond=C5=99ej Sur=C3=BD" <ondrej@sury.org> = wrote: > > > > if you run the daemon in any environment that is resource starved and > > the process(es) gets swapped than anything will be slow on first reques= t > > after period of inactivity. Not just NSD and not just any DNS server, > > but anything... > I read somewhere that some people use a cron script to send a request to = the > webserver every so often. Is this something that everybody does but that > I've somehow missed? Does it have a name? > Thanks! > O.D. Hi,=20 we have quite a few unix servers around and I am unfamiliar with a cron-set= up just to keep some sort of connectivity going.=20 First I would check the /etc/resolv.conf, if you want post it.=20 Then what I would do in your case is open a few ssh sessions, run top with = cpu focus in one, top with IO focus in another and in a third i would take = a tcpdump written to a file. Maybe another session with vmstat to check on = pagin/pageout, etc.=20 Then I'd wait the appropriate amount of time and try a web connection, if i= t takes long, I'l check the top's if there is something going on (lot's of = CPU, lot's of IO, pages etc.) and check the tcpdump in wireshark to see if = there are problematic DNS queries which maybe are timing out, etc.=20 That should cover the bases ... Maybe you are on some sort of cheap VPS ser= vice which has resources severely over-commited and maybe your whole system= has to "wake up" ... but this is a very far-fetched scenario.=20 Best=20 Ray=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?369199529.123430074.1428656504549.JavaMail.zimbra>