rpcClient should attempt a reconnect if the call fails
with 'rpc.ErrShutdown' this is needed since at times when
the servers are taken down and brought back up.
The hijacked connection from net.Dial is usually closed.
So upon first attempt rpcClient might falsely indicate that
disk to be down, to avoid this state make another dial attempt
to really fail.
Fixes#3206Fixes#3205
Do not attempt to fetch volume/drive information for
each i/o situation. In our case we do this in all calls
`posix.go` this in-turn created a terrible situation for
windows. This issue does not affect the i/o path on Unix
platforms since statvfs calls are in the range of micro
seconds on these platforms.
This verification is only needed during startup and we
let things fail at a later stage on windows.
For command line arguments we are currently following
- <node-1>:/path ... <node-n>:/path
This patch changes this to
- http://<node-1>/path ... http://<node-n>/path
These messages based on our prep stage during XL
and prints more informative message regarding
drive information.
This change also does a much needed refactoring.
Some filesystems do not implement a way to provide total inodes available, instead inodes
are allocated based on available disk space. For example CephFS, StoreNext CVSFS, AzureFile
driver. Allow for the available disk to be separately validate and we will validate inodes
only if the total inodes are provided by the underlying filesystem.
Fixes#2364
Previously checkDiskFree() checks for free available space. This
patch enables checkDiskFree() also checks for free inodes in linux and
free clusters in windows.
Fixes#2075
* Make ioErrCount updates go-routine safe
* Made ioErrCount int32 instead of *int32
... and implemented StorageAPI on *posix as opposed to posix type.
This is consistent with the thumb-rule that if a value of a type is
modified as part of the interface implementation then we implement the
interface on pointer to that type.
AppendFile ensures that it appends the entire buffer. Returns
an error otherwise, this patch removes the necessity for the
caller to look for 'n' return on short writes.
Ref #1893
This is necessary for calculating the total storage
capacity from object layer. This value is also needed for
browser UI.
Buckets used to carry this information, this patch
deprecates this feature.
Fixes#1636
New format now generates a UUID and includes it along with
the order of disks. So that UUID is the real order of disks
and on command line user is able to specify disks in any order.
This pre-dominantly solves our dilemma.
```
{
"format" : "xl",
"xl" : {
"version" : "1",
"disk": "00e4cf06-5bf5-4bb5-b885-4b2fff4a7959",
"jbod" : [
"00e4cf06-5bf5-4bb5-b885-4b2fff4a7959",
....
"c47d2608-5067-4ed7-b1e4-fb81bdbb549f",
"a543293e-99f1-4310-b540-1e450878e844",
"18f97cbe-529a-456a-b6d4-0feacf64534d"
]
},
"version" : "1"
}
```
* posix: Avoid using getAllVolumeInfo() in getVolumeDir()
This is necessary compromise to avoid significant slowness this
causes under load. The compromise is also substantial in a way
so that to avoid penalizing common cases v/s special cases.
For buckets with Caps on Unixes, we filter buckets based on the
latest anyways, so this is completely acceptable.
* XL/fs: Change the usage of verification of existance of buckets.
Optimize calling isBucketExists, it is not needed for all call
paths. isBucketExist should be called only for calls which use
temporary volume location for operations, for the rest rely on
the errors returned on their original call path.
Remove usage of filtering as well across all volume names.
This patch brings in the removal of debug logging altogether, instead
we bring in the functionality of being able to trace the errors properly
pointing back to the origination of the problem.
To enable tracing you need to enable "MINIO_TRACE" set to "1" or "true"
environment variable which would print back traces whenever there is an
error which is unhandled or at the handler layer.
By default this tracing is turned off and only user level logging is
provided.
Since there is a good amount of overlap, each code has to lock
properly for the operation they are going to perform.
- MakeVol create vols in a routine on all disks, hold locks.
- DeleteVol delete vols in a routine on all disks, hold locks.
- StatVol stat vols in a routine on all disks, hold locks.
Fixes#1588
Harshavardhana