BitFolk Users June 2012

users@mailman.bitfolk.com

15 participants
9 discussions

by Paul Stimpson

Hi everyone, Please can you recommend a domain registrar that won't treat me like poo and that won't force me to use their name servers so I can host my own DNS? Reasonable pricing and someone that doesn't throw up needless obstacles to leaving would be a plus. Thanks, Paul. -- Sent from my Android phone with K-9 Mail. Please excuse my brevity.

11 years

vps ping response times

by Joseph Heenan

Hi all, I got into a discussion with one of the people I'd recommended to bitfolk about ping times to VPSes (he was seeing some spikes). So I setup some graphs, but didn't see what I expected. Here's the graph for his vps: http://f8lure.mouselike.org/archived_graphs/baa.muuh.co.uk_day25.png which basically looks fine (which doesn't match what he was seeing from his own graphing, which is odd), and here's the graph for my VPS: http://f8lure.mouselike.org/archived_graphs/button.heenan.me.uk_day25.png The number of huge spikes (and some packet loss, shown red) on this surprised me. Would this kind of result be expected? Both graphs are for yesterday afternoon. Both vpses where pretty much idle as best we can tell. Cheers, Joseph

12 years, 1 month

At-risk notification for 2012-06-30 1100-2100Z

by Andy Smith

Hi, TL;DR version: There's some work being done on our provider's network on Saturday 30th June between the stated times. It shouldn't be noticeable. Longer version: On Saturday 30th June between 1100Z and 2100Z our colo provider will be undertaking some upgrade work on its core network. This work is not expected to have any impact on your BitFolk service. Most of the work will be taking place in racks not directly relating to BitFolk. At some point during the maintenance window one of the switches to which several BitFolk nodes are connected will be replaced with an upgraded model. All BitFolk nodes are connected to two separate switches and interface bonding should ensure that no interruption to service is experienced. These nodes are: curacao faustino kahlua obstler president urquell I will be online while the work is being carried out just in case anything does go wrong. Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting _______________________________________________ announce mailing list announce(a)lists.bitfolk.com https://lists.bitfolk.com/mailman/listinfo/announce

12 years, 1 month

IPv6 connectivity was broken between about 2105Z and 2120Z on some servers

by Andy Smith

Hi, Between approximately 2105Z and 2120Z tonight there was a total loss of IPv6 connectivity for customers on the following servers: cosmo dunkel kahlua obstler urquell This was because our transit provider did some (planned) maintenance involving switching off a router, and the above servers were not correctly set up to have a redundant IPv6 default gateway. Connectivity was restored when alerts were received and a working IPv6 default gateway was configured. It appears that the incorrect settings were used when the above servers were switched over to use interface bonding two weeks ago, to provide layer 2 resilience. Unaffected: barbar bellini curacao faustino kwak The correct redundant IPv6 gateway is now configured on all servers. Please accept my apologies for this error, and the resulting disruption it may have caused you. Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting _______________________________________________ announce mailing list announce(a)lists.bitfolk.com https://lists.bitfolk.com/mailman/listinfo/announce

12 years, 1 month

Network issue, starting around 1819Z today

by Andy Smith

Hi, As of aprox 1819Z there have been network problems at our colo/transit provider, resulting in high packet loss. This is not specific to Bitfolk; it affects all customers that I am aware of. I've spoken to provider and they are working on it, is more info as I have it. Apologies for the disruption. Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting _______________________________________________ announce mailing list announce(a)lists.bitfolk.com https://lists.bitfolk.com/mailman/listinfo/announce

12 years, 1 month

Xen Security Advisory CVE-2012-0217 - 64-bit PV guest privilege escalation vulnerability

by Andy Smith

Hi, As you may be aware from: http://seclists.org/oss-sec/2012/q2/500 there is a rather serious Intel CPU bug which can allow 64-bit paravirtualized guests to crash and/or potentially gain control of the entire node. TL;DR version: BitFolk is not affected. You don't need to do anything. Longer version: As luck would have it, BitFolk does not happen to support 64-bit guests, so the vast majority of customer systems cannot make use of this exploit. There is no technical barrier to prevent customers installing their own 64-bit OS, however, and two customers have done so. In the very short term, with the exception of the two customers who are already running 64-bit, I have prevented 64-bit guests from being booted. I shall contact the two customers individually to arrange for them to be moved to an upgraded node. The alternative would be to upgrade and reboot every node. For some nodes this would mean a full OS upgrade. This will be done eventually but over a longer time span, as it will obviously be necessary in order to support 64-bit guests. Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting _______________________________________________ announce mailing list announce(a)lists.bitfolk.com https://lists.bitfolk.com/mailman/listinfo/announce

12 years, 1 month

Re: [bitfolk] 64-bit support

by Mat Johns

>- 64-bit is alleged to be a little slower and use more memory > per-process compared to 32-bit. +1. I would have presumed that a large chunk of BitFolk's customer base are hobbyist's, developers and small businesses who are at the lower end of the RAM range. Admittedly we shouldn't make it too hard for Andy to drum up new business (as with success potentially comes more free upgrades), but for other suppliers I have dealt with who support both, I have been very particular that I wanted 32bit for the reasons we are discussing. > Even if I had a 2G RAM server, I'd still want a 32-bit system. +1. Unless you are in spitting distance of the 32bit limit, I would prefer to stick to 32bit. >> >A future memory upgrade (not planned as to when) will obviously push >> >more people into the space where>2G RAM per process would be >> >beneficial, so I will consider switching to only 64-bit by that >> >time. Will anyone be terribly upset to leave 32-bit behind? Upset, a little; no where near jumping ship (as BitFolk has always exceeded expectations) but would prefer to keep the 32bit option (if possible). Is it not possible to boot a 32bit Dom1 kernel from a 64bit Dom0? Or is the overhead of maintaining the two sets of bootstraps? ~Mat

12 years, 1 month

64-bit support

by Andy Smith

Hello, As you may be aware, BitFolk currently doesn't support 64-bit guests. The reasons for this are: - 64-bit is alleged to be a little slower and use more memory per-process compared to 32-bit. - 99.04% of BitFolk customers have less than 2GiB addressable RAM. - Supporting only 32-bit or only 64-bit is a bit simpler. We do, however occasionally lose custom due to not supporting 64-bit guests. A future memory upgrade (not planned as to when) will obviously push more people into the space where >2G RAM per process would be beneficial, so I will consider switching to only 64-bit by that time. Will anyone be terribly upset to leave 32-bit behind? To clarify I am saying that after that point new installs would have to be 64-bit. Existing 32-bit installs would be unaffected. Also if you have a requirement for 64-bit it would be appreciated if you could vote for: https://tools.bitfolk.com/redmine/issues/27 Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting

12 years, 1 month

Re: [bitfolk] hardware problems on barbar, 1826Z and ongoing

by Andy Smith

Hi Peet, On Mon, Jun 04, 2012 at 10:29:35AM +0200, Peet Grobler wrote: > Thank you for your decent, very verbose explanations of what is going wrong. I believe this distills trust in your clients. First off, we know you know what you're doing. Second - we are informed of problems, with much detail, and along the process of fixing it we are informed of what you've done so far, and what to expect. Thanks to you and others who have commented. Especially those who were affected, for your understanding. Thanks must also go to Miah Gregory who was present in our IRC channel and offered advice on how to best recover, for several hours. I was on basically the right track but Miah's advice sped things up quite a bit. Anyway this is not the last word on this matter of course. I'm still working on a plan of action and I will get back to you in due course. This is going to include service credit for those who were affected, as the data loss (even minor as it appears to have been) that arose from this sort of event was completely unacceptable and should be avoidable. > I assume you don't back up the VPSes? It's the customer's responsibility? I have no problem with this, just wondering. That is correct. We do offer to run backups for people for free, except for the disk space required to keep them in: http://bitfolk.com/customer_information.html#toc_2_Local_backups Naturally I would urge everyone to have some sort of backup regime as calamities such as human error, multiple simultaneous disk failure, datacentre fires, natural disasters, etc. do happen. Cheers, Andy -- http://bitfolk.com/ -- No-nonsense VPS hosting

12 years, 1 month

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

BitFolk Users June 2012