Netserve Service Status

15th March 2007
Minor maintenance will begin this evening (15th March) at 8pm and ending at 11pm to install a number of new servers, Cisco switches and KVM equipment. Some network restructuring may be required which could result in inconsistent network traffic for several seconds.

18th March 2007
Our final transfer of equipment and network from IFL1 will take place on Sunday 18th March. Our primary transit feeds will be run in parallel from both IFL1 and IFL2 during the day to ensure that we have a smooth transition to the IFL2 routers.

We do not anticipate significant network disruption for hosting / email service customers.

ADSL customers will be migrated from routers in IFL1 to IFL2 which may result in their ADSL sessions needing to be restarted.

20th March 2007
Mcix have advised us that maintenance on their network will take place on 20th March between 00:00 and 04:00 to replace a faulty switch within IFL1. We do not anticipate that this work will have any impact on our network.

We have been notified of the following maintenance window which may imact on our network.

Notice of maintenance window on MCIX (the IFL network) : 00:00 - 04:00
Thursday 15th March 2007

This maintenance window is required to continue diagnostic work
following previous maintenance windows which have revealed that in
certain circumstances one of the 10GE interfaces at IFL1 stops
transmitting packets.

Interruptions during this maintenance window will be limited to topology
changes as the interface is enabled and possible spanning tree problems
caused by the inteface failing to transmit spanning tree packets. The
Loop Guard feature has been enable on our switches which will help to
minimise any service interruptions.

We have suffered from a high load on our Mail storage server since early Monday morning due to a large volume of backscatter (http://en.wikipedia.org/wiki/Backscatter)

The mail indexes were rebuilt on Monday due to suspected corruption but the high loads persisted through to Tuesday when it became apparent that many clients were downloading pop mail multiple times without deleting it.

At no point in time has any of our maintenance over the last 3 days resulted in the UID lists being rebuilt, ie we report exactly what mail is in your account using the same UID's as we did last week, and the week before, etc. It is not clear why so many users mail clients decided to redownload mail.

We are advising clients to store mail on our servers for only 3 to 7 days when using pop for 2 reasons. Should you mail client download mail in full then it will not bombard you with mail or cause our servers to become overloaded. Pop accounts have limited disk quotas so storing mail for more than a week is likely to lead to over use charges being raised by ourslves to your account.

As of 9am this morning the servers were all running with normal loads and no remaining bugs. If you are aware of any remaining issues with your account please contact us via the normal support mail address.

We are trying to diagnose an upstream issue from one of our transit providers. Initially presenting to our monitoring suite as a router failure, it is now more apparent looking at traffic flows that the issue may be with the MCiX peering network. This is beyond our network but are pursuing the issue. This will be effecting connections that are persistent such as FTP IMAP and for virtual server users, SSH.

A switch fault has been encountered and dealt with this afternoon resulting in packet loss and high system loads. This has been remedied, and we are endeavoring to work around this until the issue can be fully rectified.