Maintenance work in our London suites on 23rd June [completed]

Jun 22, 2011 | Outages, Planned Work

The below work is now complete. There was a short outage in both SOV and HEX we believe all servers are now working as they are. If you have any issues please contact support.

We are planning on some preventative maintenance work to be undertaken on various items of equipment in our suites at Harbour Exchange Square (HEX) and Sovereign House (SOV) in London.

The works planned are as follows.

At SOV:

(1) Install new UPS batteries as the existing ones are reaching the end of their planned life. This operation is designed to be performed whilst the UPS is running. It should not therefore affect the electrical supply to this cabinet. However, we have a second, spare, tested UPS on-site in case anything does break on the first one and we then lose power.

Services or items that could be affected, if this UPS work fails and we have to fall-back to the spare, lasting for a period of no more than 30 minutes are:

– RADIUS authentication – we have a backup server running in another location, so ADSL users who are not connected before and actually try and connect at this time may find that authentication takes a few seconds longer. No existing connections should be lost or dropped.

– One backup name server will drop off the Merula network; there are 3 others in separate locations. There should be no effect on any clients or other services.

– One server running shared web-sites. All clients on this server who might therefore be affected have already been advised of the work planned.

– One co-lo server. This client has also been advised.

– Merula web-sites: the NOC server, portal server and mymerula sites would all be off-line for this period. The new off-site NOC box mentioned before isn’t due now to go live until early next week as the other ISP is undertaking planned works and can’t go live until then.

(2) Swap out an internal firewall. This will be done at the same time as the above work. The only sites affected will be internal Merula servers. No customer boxes or services are behind this firewall.

(3) Un-rack a couple of obsolete servers. None of these are currently live. No effect on anybody inside or outside the Merula network.

At HEX:

(1) Install a new backup router. No effect on any services or customers. It will be enabled at some future point which we’ll advise in advance.

(2) Replace cabinet UPS batteries and install a new monitoring NIC. Again, this is a hot-swap operation. The UPS should not go down. If problems do arise, again we have a spare, tested UPS on-site and maximum down-time for the services below will be 30 minutes.

– Internal monitoring server and fax server. No external effects.

– Two co-located client servers. These clients have already been advised of this possible down-time.

– One core switch – if power is lost, there will be a small amount of packet loss for 5-10 seconds as the ring stabilises. There is no direct affect on ADSL and connectivity services although there may be a few brief periods where access and traffic may be intermittent as switches & routers pickup new routing due to the work.

If you have not received an email from us, then you should expect to see nothing more, at the worst, than a little bit of instability on your ADSL connection for a period of no more than 30 minutes.

Please do email support@merula.net if you have any concerns or issues about the scope of this work.

UPDATE: Additional support routes

Jun 16, 2011 | Update

As many of you know, this site is currently the main route for notifications of outages or issues (whether planned or not).

This won’t change but we are now making a number of additional improvements to ensure that updates are available as soon as possible and, importantly, are available at all times.

The first stage in this work is to re-locate the noc server to a dual-homed location in Manchester so that (nuclear attack aside) this will be immune to any issues affecting Huntingdon and Merula and will ensure people can receive updates even if the data centre itself is unavailable for whatever reason.

We know that last night the noc server was down for a while during the outage, which, whilst planned for and announced, is less than ideal. In future, this won’t be the case.

We’ll also be changing slightly the layout, removing obsolete links and making the site easier & quicker to navigate.

There’s a very useful option available on the noc site to sign-up for an RSS feed or (shortly) an email notification which will mirror our postings here and we strongly suggest that all of our customers take advantage of this facility.

We’ll also be adding a feed from BT of their notifications to us of work & outages which will be displayed there, which will allow customers to check if this is a local, major BT or Merula related issue. This is currently available here but soon will be part of a “dashboard” of notifications on the main page.

We’ve setup a Twitter account to be fed by (or feed to) the noc site, so you can follow that account as well http://twitter.com/MerulaSupport.

We’ll also be continuing to update the telephone support number status message which should be used as the primary support route.

As a reminder to everybody, the normal support number is 0845 330 0666.

Finally, we’ve today setup a second, fallback support number which rings directly at our support centre, to be used in the event that the phone system here (and support number which routes from here) is unavailable for whatever reason. Please take a note of this new number: 0844 2011771

Maintenance work on Merula at Huntingdon 15/6/11 11pm

Jun 15, 2011 | Planned Work

At 11pm tonight, we intend making a small config change to the UPS system acting as the backup for one of our more important racks here in Huntingdon. Whilst moving this from bypass should not affect the rack in anyway, we are however advising customers that there may be a window of downtime of a maximum of 30 minutes if this fails to switchover correctly or causes other cascading issues.

Services that could be affected:

Merula website & associated sites.
A small number of servers hosted in Huntingdon.

To make sure that this is the maximum downtime, we have current backups of the configs of the core routers here as well as suitably specified and tested fall-over units available to use should any of the live ones fail to come back up for any reason. Again, we have no reason to expect this to be a problem but want to ensure that worst case is prepared for.