Commit Graph

254 Commits

Author SHA1 Message Date
Harshavardhana 9f7044aed0
fix: ignore transient errors in read path (#18006)
Errors such as

```
returned an error (context deadline exceeded) (*fmt.wrapError)
```

```
(msgp: too few bytes left to read object) (*fmt.wrapError)
```
2023-09-11 15:29:59 -07:00
Aditya Manthramurthy 1c99fb106c
Update to minio/pkg/v2 (#17967) 2023-09-04 12:57:37 -07:00
Harshavardhana 3995355150
avoid repeated large allocations for large parts (#17968)
objects with 10,000 parts and many of them can
cause a large memory spike which can potentially
lead to OOM due to lack of GC.

with previous PR reducing the memory usage significantly
in #17963, this PR reduces this further by 80% under
repeated calls.

Scanner sub-system has no use for the slice of Parts(),
it is better left empty.

```
benchmark                            old ns/op     new ns/op     delta
BenchmarkToFileInfo/ToFileInfo-8     295658        188143        -36.36%

benchmark                            old allocs     new allocs     delta
BenchmarkToFileInfo/ToFileInfo-8     61             60             -1.64%

benchmark                            old bytes     new bytes     delta
BenchmarkToFileInfo/ToFileInfo-8     1097210       227255        -79.29%
```
2023-09-02 07:49:24 -07:00
Harshavardhana 18b3655c99
with xlv2 format we never had to fill in checksumInfo() (#17963)
- this PR avoids sending a large ChecksumInfo slice
  when its not needed

- also for a file with XLV2 format there is no reason
  to allocate Checksum slice while reading
2023-09-01 13:45:58 -07:00
Harshavardhana 8a57b6bced
use renameat2 Linux extension syscall (#17757)
this is a faster and safer alternative
on newer kernel versions.
2023-08-27 09:57:11 -07:00
Harshavardhana 124e28578c
remove strict persistence requirements for List() .metacache objects (#17917)
.metacache objects are transient in nature, and are better left to
use page-cache effectively to avoid using more IOPs on the disks.

this allows for incoming calls to be not taxed heavily due to
multiple large batch listings.
2023-08-25 07:58:11 -07:00
Harshavardhana c45bc32d98
skip disks under scanning when healing disks (#17822)
Bonus:

- avoid calling DiskInfo() calls when missing blocks
  instead heal the object using MRF operation.

- change the max_sleep to 250ms beyond that we will
  not stop healing.
2023-08-09 12:51:47 -07:00
Harshavardhana 81be718674
fix: optimize DiskInfo() call avoid metrics when not needed (#17763) 2023-07-31 15:20:48 -07:00
Harshavardhana e7b60c4d65
Add slow drive timeouts to match with active disk monitoring (#17701)
allow active disk-monitoring to be configurable, and use
these add deadlines in various call layers for various
syscalls.
2023-07-25 16:58:31 -07:00
Harshavardhana a566bcf613
treat 0-byte objects to honor same quorum as delete marker (#17633)
on unversioned buckets its possible that 0-byte objects
might lose quorum on flaky systems, allow them to be same
as DELETE markers. Since practically speak they have no
content.
2023-07-11 21:53:49 -07:00
Kaan Kabalak f64d62b01d
Fix style of logOnceIf calls w/unique identifiers (#17631) 2023-07-11 13:17:45 -07:00
Poorna e8c98c3246
Avoid extra GetObjectInfo call in DeleteObject API (#17599)
Optimize DeleteObject API to avoid extra 
GetObjectInfo call on the replicating side.

For receiving side, it is just a regular
DeleteObject call.

Bonus: Fix a corner case where version purged is 
absent on target (either due to replication not yet
complete or target version already deleted in a
one-way replication or when replication was disabled). 

In such cases, mark version purge complete.
2023-07-10 07:57:56 -07:00
Klaus Post ff5988f4e0
Reduce allocations (#17584)
* Reduce allocations

* Add stringsHasPrefixFold which can compare string prefixes, while ignoring case and not allocating.
* Reuse all msgp.Readers
* Reuse metadata buffers when not reading data.

* Make type safe. Make buffer 4K instead of 8.

* Unslice
2023-07-06 16:02:08 -07:00
Kaan Kabalak 21fbe88e1f
Print certain log messages once per error (#17484) 2023-06-24 20:29:13 -07:00
Harshavardhana 1f8b9b4bd5
fix: do not listAndHeal() inline with PutObject() (#17499)
there is a possibility that slow drives can actually add latency
to the overall call, leading to a large spike in latency.

this can happen if there are other parallel listObjects()
calls to the same drive, in-turn causing each other to sort
of serialize.

this potentially improves performance and makes PutObject()
also non-blocking.
2023-06-24 19:31:04 -07:00
Harshavardhana 02c2ec3027
skip onlineDisks with parity mismatch (#17478) 2023-06-20 13:18:24 -07:00
Aditya Manthramurthy 5a1612fe32
Bump up madmin-go and pkg deps (#17469) 2023-06-19 17:53:08 -07:00
Harshavardhana 1443b5927a
allow quorum fileInfo to pick same parityBlocks (#17454)
Bonus: allow replication to proceed for 503 errors such as
with error code SlowDownRead
2023-06-18 18:20:15 -07:00
Harshavardhana 64de61d15d
fallback on etags if they match when mtime is not same (#17424)
on "unversioned" buckets there are situations
when successive concurrent I/O can lead to
an inconsistent state() with mtime while the
etag might be the same for the object on disk.

in such a scenario it is possible for us to
allow reading of the object since etag matches
and if etag matches we are guaranteed that we
have enough copies the object will be readable
and same.

This PR allows fallback in such scenarios.
2023-06-17 19:18:20 -07:00
Harshavardhana ad4e511026
do not save plain-text ETag when encryption is requested (#17427)
fixes an issue under bucket replication could cause
ETags for replicated SSE-S3 single part PUT objects,
to fail as we would attempt a decryption while listing,
or stat() operation.
2023-06-15 12:43:26 -07:00
Harshavardhana 54e544e03e
allow lookup()/head() operations on Veeam SOS objects (#17331) 2023-06-01 15:26:26 -07:00
Harshavardhana ef54200db7
offline drives more than 50% of total drives return error (#17252) 2023-05-23 07:57:57 -07:00
Krishnan Parthasarathi 3e128c116e
Add lifecycle event source to audit log tags (#17248) 2023-05-22 15:28:56 -07:00
Klaus Post aaf1abc993
simplify HardLimitReader by using LimitReader for internal usage (#17218) 2023-05-16 13:14:37 -07:00
Poorna e07c2ab868
Use hash.NewLimitReader for internal multipart calls (#17191) 2023-05-12 11:19:08 -07:00
Klaus Post 7fad0c8b41
Remove checksums from HTTP range request, add part checksums (#17105) 2023-04-28 08:26:32 -07:00
Krishnan Parthasarathi e7cac8acef
Add tags to auditLogLifecycle (#17081) 2023-04-26 17:49:00 -07:00
Praveen raj Mani 72802a5972
Use 'minio/pkg/sync/errgroup' and 'minio/pkg/workers' (#17069) 2023-04-25 22:57:40 -07:00
Harshavardhana b1f3935c5b
allow ListObjects() when a prefix is an object (#17074) 2023-04-25 22:41:54 -07:00
Krishnan Parthasarathi fae9000304
heal: Pick maximally occuring modTime in quorum (#17071) 2023-04-25 10:13:57 -07:00
Harshavardhana 84f31ed45d
simplify MRF, converge it to regular healing (#17026) 2023-04-19 07:47:42 -07:00
Harshavardhana 6825bd7e75
fix: inlined objects don't need to honor long locks (#17039) 2023-04-17 12:16:37 -07:00
Klaus Post c133979b8e
Add part count to checksum (#17035) 2023-04-14 09:44:45 -07:00
Poorna cd6dec49c0
Add trace support for ilm activity (#16993) 2023-04-11 19:22:32 -07:00
Harshavardhana c06e0bfef9
set correct `Host:` value for replication event notification (#16984) 2023-04-06 10:20:53 -07:00
Poorna 699a24f7e5
batch: validate versioning on src/tgt buckets (#16955) 2023-04-04 10:50:11 -07:00
Harshavardhana 216a471bbb
on quorum DeleteObject() errors attempt an MRF (#16932) 2023-03-31 08:15:41 -07:00
Harshavardhana 6c11dbffd5
add crash protection from backend modifications (#16846) 2023-03-20 09:08:42 -07:00
Poorna d1e775313d
support decommissioning of tiered objects (#16751) 2023-03-16 07:48:05 -07:00
Harshavardhana e0f4dd6027
remove unncessary logs from WalkDir(), PutObject() (#16818) 2023-03-15 11:52:23 -07:00
ferhat elmas 714283fae2
cleanup ignored static analysis (#16767) 2023-03-06 08:56:10 -08:00
Klaus Post 9acf1024e4
Remove bloom filter (#16682)
Removes the bloom filter since it has so limited usability, often gets saturated anyway and adds a bunch of complexity to the scanner.

Also removes a tiny bit of CPU by each write operation.
2023-02-24 09:03:31 +05:30
Harshavardhana a0f06eac2a
add Veeam SOS API first implementation (#16688) 2023-02-22 19:54:57 +05:30
Krishnan Parthasarathi d136ac0596
Don't close transition task channel on server exit (#16627) 2023-02-15 22:09:25 -08:00
Krishnan Parthasarathi cea2ca8c8e
Add restore-status header for multipart objects (#16508) 2023-01-31 07:53:45 +05:30
Harshavardhana 67fce4a5b3
fix: dangling delete() upon success should return 404 (#16494) 2023-01-27 12:43:45 -08:00
Anis Elleuch d98116559b
Use async healing in PutObject call (#16431) 2023-01-19 00:54:22 -08:00
Harshavardhana 2937711390
fix: DeleteObject() API with versionId under replication (#16325) 2022-12-28 22:48:33 -08:00
Anis Elleuch acc9c033ed
debug: Add X-Amz-Request-ID to lock/unlock calls (#16309) 2022-12-23 19:49:07 -08:00
Krishnan Parthasarathi 2fa35def2c
Fix DeleteObject when only free versions remain (#16289) 2022-12-21 16:24:07 -08:00