fix: stop full database rewrites on every plugin db call

This commit is contained in:
LukeGus committed 2026-10-05 20:06:32 -05:00
1 parent 40df00bd95
commit 7508a45f36
14 files changed
+322 -69

No files matched your search

+6 -40
View File
@@ -1,58 +1,24 @@
<!-- SUMMARY -->
A patch for 2.9.0 that fixes OIDC sign-in after upgrading, Docker and Host Metrics logins, memory and CPU use, file downloads, tmux sessions, user data export and more. Also adds predefined host tags.
An urgent patch for 2.9.1 that fixes the server writing huge amounts of data to disk, which could add up to 100 GB every few hours. Update as soon as you can.
<!-- /SUMMARY -->
<!-- YOUTUBE -->
https://youtu.be/lngaePO96tM
<!-- /YOUTUBE -->
<!-- UPDATE_LOG -->
- Added predefined host tags that admins can set, with suggestions in the host editor
- Kept compact host row actions on the same line as the host name
- Added a reload button when a sign-in verification method fails to load
- Admins can now merge a duplicate SSO account made by 2.9.0 back into the original account
- Hosts behind the same jump hosts now share one connection for status checks
- Simplified host status to just online or offline, removing the "last login failed" state
- Host status now updates every 15 seconds without a page refresh, and status checks run every 30 seconds by default
- Completed the Simplified Chinese translation
- Documented cookie and API key sign-in in the API docs
- Background database saves now happen at most once every 30 seconds, so disk writes no longer grow with the number of hosts
- Plugin database and storage reads are now logged to the audit log once per start instead of on every read
- On first start, 2.9.2 clears the plugin read entries that 2.9.1 piled up in the audit log
<!-- /UPDATE_LOG -->
<!-- BUG_FIXES -->
- OIDC accounts from 2.8 not being able to sign in, or getting a new empty account, after upgrading
- Downloading a file in the file manager turning the whole screen black
- tmux sessions not resuming and new sessions being created instead
- Connections through two or more jump hosts failing
- Docker and Host Metrics failing to sign in while the terminal worked
- Fleets, Proxmox and automations not getting the host sudo password
- Server running out of memory on start and when importing hosts
- High CPU use after the app was first opened
- User data export freezing the server
- 1Password and other secret references being rejected as SSH keys
- Slow file deletes in the file manager
- Mac App Store app failing to start its local server
- Snippets failing to sync on older databases
- GitHub sign in failing
- RDP showing scrollbars and jittering near the edges
- Status checks triggering Fail2Ban bans while a session was open
- Too many database writes from session activity
- Host editor tabs being hidden when they did not fit
- SSH host keys being accepted without a check when the host was missing from the database
- Host status only updating after a full page refresh
- Every host showing offline right after signing in
- Host status flipping between online and offline on refresh or after one dropped packet
- Hosts showing "last login failed" when they connected fine
- Hosts behind a jump host staying offline after the jump connection dropped
- Host status being wiped in the desktop app when the host list failed to load
- Tunnels, remote desktop logins and other host settings missing from shared hosts in the desktop app
- SSO and LDAP provider dialogs overflowing the screen and using mismatched toggles
- Plugin audit log entries failing to save when no user was signed in
- The server rewriting its whole database to disk many times a second, causing very high disk writes (up to 100 GB every few hours) on 2.9.1
- The audit log filling up with plugin entries and pushing out real ones
<!-- /BUG_FIXES -->