JSWeb Server Status Feb-March = 100%

March 8th, 2010

From: Feb 01 2010 (EU/Western) – To: Mar 07 2010 (EU/Western)

1. HTML Cluster 2 Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

2. HTML Cluster 1 Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

3. Bravo MySQL/MYSQL Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

4. India (mail.jswebconsole.co.uk) POP3 Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

5. Hotel (JSWeb + Development) Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

6. FULL Dedicated Servers
- uptime: 100.000%
- downtime: 00min
- outage: 0

29th Jan 2010 1am – 30 min maintenance window

January 12th, 2010

In order to upgrade our HA (High Availability) firewall cluster we require a 30min window which will result in an outage to all services (other than DNS).

The firewall change has worked successfully but a routine up date at the same time to update the kernel on bravo – a database server in the shared hosting environment has failed and the machine is not currently restarting – engineers are investigating

Network Issue – 20/12/2009 16:12

December 20th, 2009

A network routing issue has occured which is preventing some users from connecting to the servers in our primary data centre.

All servers are running correctly.

3rd party Network Engineers are investigating the issue and are reporting back to JSWeb frequently

Update 16:50: – we are able to connect to the servers from our secondary data centre which suggests the issue is definitely with the external network carrier bringing inbound traffic to the primary centre.. engineers investigating and contacting 3rd party network carriers

Update 16:56: – external network carrier issue confirmed, they have given an initial eta to fix of 1 hour from now

Update: 1745: There appears to have been a major failure of the primary carrier in Dublin 17, and the failover to secondary routes is incomplete (currently INEX peered routes appear to be working ok).

Update: 1830: From the datacentre: Engineers are continuing to work on the issue. It appears there is an issue with routing to the Dublin 17 AS via Smart Telecom, and issues with the other carriers filtering the routes out. Smart are reporting a major problem on the T50 fiber ring, which is also impacting their ability to resolve this issue. We continue to work to resolve this serious issue and will post regular updates.

Comment: JSWeb are unable to supply any more detail

Update: 1845: From the datacentre: All services have been restored. We apologise for any impact this incident may have caused.

Email server migration 16/12/2009 11pm GMT

December 16th, 2009

All

As part of our migration of some key services away from virtual environments back into physical ones we will require a maintenance window starting from 11pm GMT on 16/12/2009 in order to migrate our email server.  During this period websites will continue to operate normally but you will not be able to send or receive email through the jsweb email server.  We expect the work to be complete by 3am GMT 17/12/2009 and obviously we will make ever effort to complete the work as soon as possible.

HOSTING ENVIRONMENTS: Changes to help your business move forward

October 24th, 2009

Date: 24-10-09
Time: 12:30

HOSTING ENVIRONMENTS: Changes to help your business move forward

Dear all cluster 1 shared hosting clients;

Over the last few weeks we have been experiencing some shared hosting problems which obviously we are not happy with!

There has been a combination of causes including:

  • An iframe attack which is widespread across the internet, not only infecting clients’ websites but possibly the PC’s of their site visitors. This is usually due to passwords, specifically FTP passwords, not being SECURE (e.g. using ‘mypassword’) or clients’ own PC’s which are infected but have not been checked (we use Avast which is quite good!)
  • Increases to some of cluster 1 clients number of visitors which is great news, but also has an impact on overall server loads (Hogging!)
  • ……AND yes we are a little disappointed with our shared hosted cloud environment.

Let me explain:

The cloud is a new technology and in principle (on paper) looks quite amazing. It has already being adopted by large data centres such as Microsoft, Amazon, Facebook etc and for our clients who have opted for DEDICATED VM’s the performance has been great!

However, the cloud is having problems with our shared environment – one technology talking with another doesn’t always work! We only have to remember Microsoft’s release of VISTA which looked and worked quite well by itself but became annoying when you found your printer or certain applications wouldn’t work with it! — well that is what we are experiencing within our shared cloud.

At JSWeb, we pride ourselves in being eCommerce specialists, and we have had to make a decision. Ideally, any eCommerce site – or any that is serious about making money or where it is the owner’s sole or main source of income, should be in it’s own dedicated environment where there is no risk of interruption to business by the activities of another client. However, the cost of a Managed Dedicated VM (just £6 per day) is deemed too expensive by many of our current cluster 1 clients, so to move us FORWARD (and become more reliable) we have decided to move our shared hosting performance environment back to PHYSICAL hardware.

We are in full discussions with our suppliers (H365 and IBM) to get our new equipment built and plan a migration from cluster 1 to the new physical boxes with minimal interruption. We will keep you updated.

Again all we can do is apologise – unfortunately problems will never fully disappear within shared hosting as you will always be at the mercy of the actions of others.

Regards

Jamie
Owner of JSWeb Ltd

———————-
UPDATE: 29 October 09

Summary: I know it is Halloween weekend and I don’t want to give people a SCARE but we have made a decision to move our databases out of the CLOUD environment to their new PHYSICAL SAN environment. This is one of the biggest moves and although normally we would want to give lots of notice, an opening at our NOC became available which we have grabbed allowing us to move quickly.

SCHEDULED CHANGE: Fri 30th October at 01:00 AM (Downtime 1-2hrs)

UPDATE: 31 October 09
All work was completed within the schedduled time scale and now our DB is on physical — Interested to know if anyone notices any difference of any kind

UPDATE: 16 NOV 09
The brand new HP DL380’s (9x 4 cores with 8Gb’s) have now been ordered.
Once these have been built then we will be ready to move VM’s from Cluster 1 and 2 to their new physical homes.
We have planned out a method to move each VM one by one (this is a cool feature of the cloud).

This work being scheduled aims to be completed by 27th November 2009.


PLANNED OUTAGE: 23rd Nov from 11:30 PM
Task ahead: Moving the Storage Area Network (SAN) – migrating all site data from virtual to physical machines – will require complete outage of service for several hours

  • First cluster1, then cluster 2 then email storage
  • Outcome: The SAN migration started at 23:00 last night and has successfully been completed by 07:39 this morning.