I added 8 dns names to my whitelist through the "Edit Clients" for my Server Class on my deployment server.
I saved and reloaded, then went back in to edit it.
I removed all but one entry, hit save and reload and got the following error:
In handler 'serverclasses': Gap in numbered regexes: expected attribute=whitelist.8 not found.
I believe this is a bug within the Edit Clients UI.
It sounds like a bug.
For a work-around, go to the serverclass.conf file. You will probably see entries like this somewhere in the file:
whitelist.0= yourhostname1
whitelist.1= yourhostname2
whitelist.3= yourhostname8
Or perhaps the first whitelist.x is not whitelist.0!
In serverclass.conf the whitelists are numbered. The first one must be whitelist.0 and no numbers can be skipped in the list. If you see that these rules have been violated then
./splunk diag before fixing the problem.serverclass.conf: ./splunk reload deploy-serverEven if the workaround solves the problem, please report it to http://support.splunk.com
Same problem here, had to heavily edit a whitelist and got the same kind of whining (note! new whitelist was shorter than the old one!)
I worked around this by deleting a few hosts (one by one + save) from the whitelist until I had a number of hosts = the # hosts I was adding, then it allowed me to modify the whole list.
definitely sounds like a bug
It sounds like a bug.
For a work-around, go to the serverclass.conf file. You will probably see entries like this somewhere in the file:
whitelist.0= yourhostname1
whitelist.1= yourhostname2
whitelist.3= yourhostname8
Or perhaps the first whitelist.x is not whitelist.0!
In serverclass.conf the whitelists are numbered. The first one must be whitelist.0 and no numbers can be skipped in the list. If you see that these rules have been violated then
./splunk diag before fixing the problem.serverclass.conf: ./splunk reload deploy-serverEven if the workaround solves the problem, please report it to http://support.splunk.com
I had a whitelist that was number starting from 1 rather than 0. I never saw any messages about it in splunk 5.1.3, but after I upgraded to 6.1.5 I did. However though it told me the correct server class, it incorrectly identified the problem as missing whitelist.9 rather than whitelist.0 #headscratcher (until I read this post).
I filed a bug report: link text