Hello,
Recently my VPS has begun performing certain IO intensive operations
notable slower, in some cases maildirs that can take less than a second
to open under ideal circumstances have been taking up to a minute to
load. I am told by Andy that iowait on the host server (barbar) is not
high at the moment, so it is likely to be something to do with the VPS
itself.
Can anyone suggest any diagnostic measures I can take to identify where
the performance issue is? I have never had to deal with such a
situation before, and Google is not offering a significant amount of
guidance.
Thanks,
Robert
Hi there
I'm running Ubuntu 8.04 on the VPS (image provided by Bitfolk) and periodically the VPS seems to lock up and I get the following scrolling really fast on the console:
[1366226.746076] =======================
[1366238.351907] BUG: soft lockup - CPU#0 stuck for 11s! [apache2:27046]
[1366238.351920]
[1366238.351925] Pid: 27046, comm: apache2 Tainted: G D (2.6.24-24-xen #1)
[1366238.351929] EIP: 0061:[<c03285fa>] EFLAGS: 00200286 CPU: 0
[1366238.351938] EIP is at _spin_lock+0xa/0x10
[1366238.351940] EAX: d6002dd8 EBX: d6002d94 ECX: c156a9c0 EDX: 00000000
[1366238.351943] ESI: c156a9c0 EDI: 00000000 EBP: d4a4bdd8 ESP: d4a4bc3c
[1366238.351946] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0069
[1366238.351954] CR0: 80050033 CR2: 08e080c8 CR3: 04d26000 CR4: 00002620
[1366238.351958] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[1366238.351961] DR6: ffff0ff0 DR7: 00000400
[1366238.351965] [<c01a6ebd>] try_to_free_buffers+0x2d/0x90
[1366238.351978] [<c0167c75>] shrink_page_list+0x4c5/0x600
[1366238.351985] [<c0166e0f>] isolate_lru_pages+0x5f/0x1c0
[1366238.351991] [<c0167ecf>] shrink_inactive_list+0x11f/0x3b0
[1366238.351998] [<c01681fc>] shrink_zone+0x9c/0x100
[1366238.352002] [<c0168d42>] try_to_free_pages+0x152/0x250
[1366238.352008] [<c0162f9b>] __alloc_pages+0x14b/0x390
[1366238.352013] [<c011959c>] kmap_atomic+0x1c/0x30
[1366238.352020] [<c01713dc>] handle_mm_fault+0x8ac/0x1360
[1366238.352025] [<c0164cf2>] test_clear_page_writeback+0x72/0x100
[1366238.352030] [<c0275361>] __add_entropy_words+0x61/0x1f0
[1366238.352037] [<c011e34e>] __wake_up+0x3e/0x60
[1366238.352042] [<c027571b>] add_timer_randomness+0xdb/0x190
[1366238.352047] [<c032a18e>] do_page_fault+0x35e/0xe70
[1366238.352052] [<c01581d1>] handle_IRQ_event+0x41/0x80
[1366238.352056] [<c0159ad6>] handle_level_irq+0xb6/0xf0
[1366238.352061] [<c0254290>] evtchn_do_upcall+0xc0/0x1a0
[1366238.352067] [<c0329e30>] do_page_fault+0x0/0xe70
[1366238.352071] [<c0328ad5>] error_code+0x35/0x40
[1366238.352077] =======================
I've increased the swapfile size to 1GB as I noticed we were using it pretty heavily. The VPS runs a couple of websites, backed by MySQL.
Any ideas?
Alex
This email carries a disclaimer, a copy of which may be read at http://learning.longhill.org.uk/disclaimer
Hello,
The most common support request is to reset BitFolk account
password. I have therefore implemented a password reset feature to
the web panel at https://panle.bitfolk.com/.
It works as you would expect:
- Someone follows the "reset password" link on the login screen.
- They are invited to put in an account name. If the account exists
and reset is enabled for that account, then an email is generated
with an authorisation key in it. The email is sent to the email
address on record for that account.
The email also contains the IP address of the client that
requested the reset, together with their browser's user agent and
the time stamp.
- Following the link in the email within 12 hours will randomly
generate a new password. Nothing will happen if the link is not
followed.
When this was last discussed, there were a few people who didn't
like the idea of their passwords being able to be reset by email and
potentially being exposed to an attacker who already has control of
their email account. Therefore I have also added the option to
disable the feature.
I will put the feature live on Wednesday 27th January, so if you
don't like the idea of the above being possible please go to
https://panel.bitfolk.com/account/security/ and disable it.
In the near future I shall also add a third option for PGP encrypted
emails, which should allay most people's concerns about that
feature, but I am keen to get a basic version of this deployed
before the weekend of 6th February as I'll be at FOSDEM and support
will be emergencies only.
Clearly forms to change other personal details such as email
and postal addresses are also required, and also will be coming
soon.
On the security link above you will also find a password change
form for those who prefer a web page vs. logging in to the Xen Shell
and typing "passwd".
As always I'm keen to hear what people's priorities are for other
features to be added to the panel. I'm also open to ideas about
software for managing such feature requests, as RT probably isn't
best suited to it.
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"Whoever is responsible for this stunning design failure deserves continuous
cockpunches." -- jwz
Me again!
I'm slightly confused about one thing. Still trying to follow the
instructions here http://flurdy.com/docs/postfix.
I trying to get my head around a slight conflict. The server is
delivering email to /var/mail/virtual/graeme. Definitely arriving. But
my mail clients appear to be looking in a different directory,
/home/graeme/Maildir.
How does it move across? Am I missing something blindingly obvious?
Many thanks,
Graeme
Hi Bitfolks,
I've been using the very comprehensive instructions I found here
http://flurdy.com/docs/postfix/ to set up my shiny new VPS box. I
'think' I've got it mostly working now, but I'm not sure how to get my
box talking to the bitfolk central spamassassin box.
I realise I have to specify a username and the servername, but I don't
know the syntax. Could someone point me in the right direction? I assume
it's just a line in master.cf I need.
Many thanks,
Graeme
Hi,
Aside from the known issues with newer Ubuntu kernels not working on
the Xen hypervisor from Debian Lenny (which is what BitFolk uses),
there also appears to be a lack of compatibility between newer
releases of Ubuntu and older kernels:
https://bugs.launchpad.net/ubuntu/+source/mountall/+bug/447747
This will leave you in the position of not being able to boot with
the supplied kernel and not being able to revert to a kernel from
Hardy or Lenny.
There may be a workaround, I'm not sure yet, but at the moment I
wouldn't recommend trying to do this unless you know you can make it
work (or know you can fix it from the rescue env.).
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"I'd be happy to buy all variations of sex to ensure I got what I wanted."
-- Gary Coates (talking about cabling)
Hi everyone,
I could do with a bit of advice.
I've set up bind with the following in named.conf.local for each domain:
----------------------------------------------------------
zone "polyvisual.co.uk" {
type master;
file "/etc/bind/db.polyvisual.co.uk";
};
----------------------------------------------------------
and the following db file for each domain:
----------------------------------------------------------
;
; BIND data file for local loopback interface
;
$TTL 86400
@ SOA ns1.polyvisual.co.uk. hostmaster.polyvisual.co.uk. (
2 ; Serial
14400 ; Refresh (4 hrs)
900 ; Retry (15 mins)
1209600 ; Expire (2 weeks)
3600 ) ; Negative Cache TTL (1 hr)
NS ns1.polyvisual.co.uk.
NS a.authns.bitfolk.com.
NS b.authns.bitfolk.com.
NS c.authns.bitfolk.com.
MX 5 mail.polyvisual.co.uk.
A 212.13.195.149
ns1 A 212.13.195.149
www CNAME polyvisual.co.uk.
mail A 212.13.195.149
----------------------------------------------------------
To make things easy, if I wanted to just use ns1.polyvisual.co.uk as the
nameserver for each domain [rather than ns1.whateverthedomainis.co.uk]
should I change the SOA to ns1.polyvisual.co.uk in each of the db files
[leaving the domain name correct in the rest of the db file?]
Cheers,
Matt
I have just started working at a new office.
I just re-installed a windows box to have Lenny on it.
The box works fine.
I noticed that my SSH sessions to computers within the company were
hanging, then eventually timing out with a reset by peer message. It is
actually afecting all networking, that's just what made me realise that
the network sometimes "goes away".
No one else is having these problems.
To make progress I bought in my laptop and tried to use that, then
suddenly found the same issue.
I have plugged my laptop into a different port, one next to someone
elses workstation with no effect.
I have since been playing with things, looking through the internet and
playing with ping.
There are no entries in syslog/messages/dmesg about the NIC or network
stack at, or around, the time when the network "disappears".
Basically I lose all connectivity in terms of send/recv but ifconfig
reports that the adapter is up and working.
Both machines are running Lenny. My laptop has been stable for
aaaaaaages, has never had problems with wired ethernet on multiple
networks it's been hooked up to...and is even running a newer kernel to
get some fixes for the wifi driver (not used, I'm on wired ethernet here)
The box I just installed is a beast of a dell server and is using a
vanilla Lenny install from CD with only a base system + Gnome so far.
I have searched the interwebs and found the advice regarding tcp window
scaling, this does not get rid of the problem on my laptop.
I'm having a hard time finding anything that seems relevant to me.
If anyone has seen this, or has an idea where to look I'd appreciate it.
n
Hi,
At approximately 1106Z today, suite TFM1 in Telehouse lost power for
around 2 minutes. The only live BitFolk server in TFM1 at present
is dunkel.
Servers are set to not automatically start every VPS, as if they have
been rebooted it is usually for maintenance purposes. What should
have happened is that I should have been alerted, and then I would
have sorted things out.
A significant part of the internal monitoring is currently running
on a VPS on dunkel, but there is external monitoring for situations
such as this.
However, this has not worked and it wasn't until I received a phone
call on the urgent support number that I was aware of the problem.
As I type this, all VPSes on dunkel have now been restarted -- some
may still be fscking.
I am waiting on Telehouse for more information on the power outage
but obviously the real failing here is one of BitFolk's monitoring.
This is the second time there has been a monitoring failure; the
system had actually been changed since the first time, although that
was a different class of failure. I will be looking into why the
external monitoring failed, and changing how it's done if necessary.
In the meantime I will be adding a day of free service to all those
on dunkel who were affected.
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"I'd be happy to buy all variations of sex to ensure I got what I wanted."
-- Gary Coates (talking about cabling)
Hi Bitfolks,
I've just signed up to Bitfolk and could do with a bit of assistance
with DNS.
I've signed up for a couple of domain names with godaddy.com. I've
bought simmo.gs and goodwithwords.eu.
I'm not certain of the order I should be doing things and the
configuration I should be setting up on my debian vps box.
Does anyone have any guidance they could offer? Or even a sample zone
file that I could copy.
I have done this sort of thing before, but it's been a while and I
appear to have forgotten swathes. I won't be offended to be treated like
a numpty!
Cheers,
Graeme
