I don’t know MogileFS, what is the rationale for this form of URL?
I was assuming that each file have the same URL path component for each
mogstred-server, so you only need to dynamically know the upstream
address and port.
I don’t know MogileFS, what is the rationale for this form of URL?
I believe the .fid is file ID. So “0000000005” is the file ID, and it
spreads the file over a seg/ment/ed directory to make sure there’s not
too many files in one dir…
I was assuming that each file have the same URL path component for each
mogstred-server, so you only need to dynamically know the upstream address
and port.
I am not sure if this is right.
I believe MogileFS will store the same file however it’s host and
“devX” that is dynamic, I need to spend more time with a multi-storage
node and double check this
AFAIK, the dev number is a completely arbitrary int and is set by mogadm when you setup the cluster.
Actually I saw a thread recently and I believe it was talking about
devX numbers - some guy had an issue where it skipped from 1, 2, 3, 5
and the code seemed to go incrementally and failed when it hit the
missing 4.
I am pretty sure it was the dev IDs…
Either way, I believe the full URL is what is needed for this to be
useful, not just a single http://a.b.c.d:port/ since the devX could be
there or not, I believe.
Do you know if there’s better documentation out there? I had to dig
into the perl source code in order to get it setup.
Nope. I’d ask the mailing list there any questions, Google, etc.
They have a wiki up but it doesn’t have much info and last I checked
the navigation sent you in a circle.