mirrors/scoutfs
1
0
Fork 0
mirror of https://github.com/versity/scoutfs.git synced 2026-01-08 21:03:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Allow replaying srch file rotation

Browse Source 
When a client no longer needs to append to a srch file, for whatever
reason, we move the reference from the log_trees item into a specific
srch file btree item in the server's srch file tracking btree.

Zeroing the log_trees item and inserting the server's btree item are
done in a server commit and should be written atomically.

But commit_log_trees had an error handling case that could leave the
newly inserted item dirty in memory without zeroing the srch file
reference in the existing log_trees item.  Future attempts to rotate the
file reference, perhaps by retrying the commit or by reclaiming the
client's rid, would get EEXIST and fail.

This fixes the error handling path to ensure that we'll keep the dirty
srch file btree and log_trees item in sync.  The desynced items can
still exist in the world so we'll tolerate getting EEXIST on insertion.
After enough time has passed, or if repair zeroed the duplicate
reference, we could remove this special case from insertion.

Signed-off-by: Zach Brown <zab@versity.com>
This commit is contained in:
Zach Brown
2023-01-17 14:22:49 -08:00
parent 01c8bba56d
commit 2e2ccb6f61
2 changed files with 16 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
kmod/src/server.c
View File

@@ -1586,6 +1586,13 @@ static int server_commit_log_trees(struct super_block *sb,
if (ret < 0 || committed)
goto unlock;
/* make sure _update succeeds before we modify srch items */
ret = scoutfs_btree_dirty(sb, &server->alloc, &server->wri, &super->logs_root, &key);
if (ret < 0) {
err_str = "dirtying lt item";
goto unlock;
}
/* try to rotate the srch log when big enough */
mutex_lock(&server->srch_mutex);
ret = scoutfs_srch_rotate_log(sb, &server->alloc, &server->wri,
@@ -1600,6 +1607,7 @@ static int server_commit_log_trees(struct super_block *sb,
ret = scoutfs_btree_update(sb, &server->alloc, &server->wri,
&super->logs_root, &key, &lt, sizeof(lt));
BUG_ON(ret < 0); /* dirtying should have guaranteed success */
if (ret < 0)
err_str = "updating log trees item";

8
kmod/src/srch.c
View File

@@ -995,6 +995,14 @@ int scoutfs_srch_rotate_log(struct super_block *sb,
le64_to_cpu(sfl->ref.blkno), 0);
ret = scoutfs_btree_insert(sb, alloc, wri, root, &key,
sfl, sizeof(*sfl));
/*
* While it's fine to replay moving the client's logging srch
* file to the core btree item, server commits should keep it
* from happening. So we'll warn if we see it happen. This can
* be removed eventually.
*/
if (WARN_ON_ONCE(ret == -EEXIST))
ret = 0;
if (ret == 0) {
memset(sfl, 0, sizeof(*sfl));
scoutfs_inc_counter(sb, srch_rotate_log);