vitalif
0 Followers · 0 Following
  • Joined on 2017-02-23
Repositories
98
 Projects Packages Public Activity Starred Repositories
vitalif pushed to fatal-error-hook at vitalif/vitastor 2026-10-28 01:26:20 +03:00
febf81be79 WIP Replace exit/abort in client with on_fatal_error_hook
f07ff4e479 Change cluster_client_t::msgr to unique_ptr
23b9bff8fb Add a rename-dir test to test_nfs
e2c5ec8824 Stop NFS prefix scans at the end of the prefix
0757a1e8b9 Add an end key to K/V listings
Compare 84 commits »
vitalif commented on pull request vitalif/vitastor#110 2026-10-28 00:24:23 +03:00
Pass the checked device nodes to the OSD in vitastor-disk exec-osd

Что собственно смущает, смущает что (допустим) если глюканёт рейд контроллер и между exec-osd и запуском osd…

vitalif commented on pull request vitalif/vitastor#110 2026-10-28 00:23:24 +03:00
Pass the checked device nodes to the OSD in vitastor-disk exec-osd

Тут х-з. Меня такой вариант смущает. Кейс под который ты это фиксишь очень узкий, запуск OSD поверх dm-multipath это…

vitalif merged pull request vitalif/vitastor#109 2026-10-28 00:17:46 +03:00
Set ownership of OSD partitions in a udev rule of its own
vitalif pushed to master at vitalif/vitastor 2026-10-28 00:17:46 +03:00
bbcfe3e258 Set ownership of OSD partitions in a udev rule of its own (#109)
vitalif commented on pull request vitalif/vitastor#114 2026-10-28 00:15:20 +03:00
Store v1: Fix two flusher hangs with deletions

Мне тут не нравится вот этот фикс

  1. Store v1: Do not lose the flush of a write over a deletion

The write is renumbered from version 1 once the deletion is gone,…

vitalif merged pull request vitalif/vitastor#112 2026-10-28 00:10:45 +03:00
NFS: Do not memcpy from NULL when encoding an empty opaque
vitalif pushed to master at vitalif/vitastor 2026-10-28 00:10:45 +03:00
c0d6a417ba NFS: Fix UBSan memcpy from NULL error (#112)
vitalif closed pull request vitalif/vitastor#111 2026-10-28 00:10:01 +03:00
Stop K/V listings at the end of their range
vitalif commented on pull request vitalif/vitastor#111 2026-10-28 00:10:00 +03:00
Stop K/V listings at the end of their range

Замержил без этих тестов, но в test_nfs добавил тест на переименование поверх пустой директории

vitalif pushed to master at vitalif/vitastor 2026-10-28 00:09:36 +03:00
23b9bff8fb Add a rename-dir test to test_nfs
e2c5ec8824 Stop NFS prefix scans at the end of the prefix
0757a1e8b9 Add an end key to K/V listings
Compare 3 commits »
vitalif commented on pull request vitalif/vitastor#108 2026-10-27 23:56:13 +03:00
Exit with code 1 instead of aborting when the OSD fails to open its devices

Замержил без теста

vitalif closed pull request vitalif/vitastor#108 2026-10-27 23:56:13 +03:00
Exit with code 1 instead of aborting when the OSD fails to open its devices
vitalif pushed to master at vitalif/vitastor 2026-10-27 23:56:06 +03:00
cfaa3feefa Exit with code 1 instead of aborting when the OSD fails to open its devices
vitalif merged pull request vitalif/vitastor#113 2026-10-27 23:53:59 +03:00
Fix a signed shift overflow in sha256_transform()
vitalif pushed to master at vitalif/vitastor 2026-10-27 23:53:59 +03:00
4e638b2681 Fix UBSan signed shift overflow in sha256_transform() (#113)
vitalif pushed to fatal-error-hook at vitalif/vitastor 2026-10-27 10:28:06 +03:00
a7d9daa5bd WIP Replace exit/abort in client with on_fatal_error_hook
vitalif commented on pull request vitalif/vitastor#107 2026-10-27 00:30:40 +03:00
Store v2: Wait for fsyncs submitted by other operations before completing a sync

тут точно обе правки нужны? выглядит как будто то ли два разных кейса поправлено, то ли один, но двумя способами. верхней правки не хватит ли?

vitalif commented on pull request vitalif/vitastor#106 2026-10-27 00:28:12 +03:00
osd: Roll back unstable writes of removed pools in the old store

тут два раза чтоб не делать pgs =, можно выше pgs = 1 раз присвоить а тут только проверять

vitalif commented on pull request vitalif/vitastor#106 2026-10-27 00:28:12 +03:00
osd: Roll back unstable writes of removed pools in the old store

тут делается rollback_(), а выше только check_(). как будто не совсем сбалансированно, надо по идее одно и то же триггерить, ну либо в одном месте где-то триггерить