Simply because of the size and age of the environment, some of our production clusters have now reached the limit (including local paths) – you see this message in the logs
[2012-08-20 01:48:52.256 77C3DB90 info ‘ha-eventmgr’] Event 2003 : The maximum number of supported paths of 1024 has been reached. Path vmhba3:C0:T4:L0 could not be added.
In this state the hosts drop SAN LUNs because they can’t drop the locally attached storage, even though we are not using it. You can see how many paths are being used on a specific host by selecting the host, going to Configuration > Storage Adapters and selecting the local storage adaptor:
In fact you can see that this IBM blade is registering two local paths, one for the storage and one for the ServeRAID and one for the disk enclosure. Removing some LUNs reduced the number of paths and, now that it’s fixed, if I look at the Fibre Channel HBAs you can see that each one has 126 LUNs and 4 targets making 504 paths, there are two Fibre HBAs and 2 local paths making a total of 1010 paths (504*2 + 2 = 1010).
Before we removed some LUNs, one adaptor had 512 LUNs and the other 510 – if you add the local adaptor paths that makes you 1024 paths (512 + 510 + 2 = 1024). This meant that the LUN in the error message had been dropped to satisfy the limit.
Of course, you want to avoid being this close to the limit: but if like me there are compelling requirements to be this close, make sure you factor in the local storage!