Changelog napp-it cs
####################

26.06i
 - import repli, snap and scrub jobs from napp-it SE (incremental replications can go on) 
 - webserver.pl modification ti limit RSS under heavy load
 - guards to restart webserver.pl and monitor.pl if RSS RAM consumption too high

26.06h
 - Windows disk detection with batch processing (eg smartmontools, faster)
 - RSS guard for webserver and monitor to limit RAM consumption due threads inheriting env
 - get-pools faster with many pools

26.06g
 - fix a problem in monitor.pl cmd caching
 - optimize resource demands in monitor and replication

 26.06f
 - fix a possible replication problem with large pools

26.06e_rc
---------------------
 - Debian 13 demands systemd
 - fix restart problem on Debian 13
 - fix a memory problem with Ajax (admin-mon used in status ample)

26.06d_rc
---------------------
- guest share fix for ksmba and samba (rw, ro)

26.06b_rc
---------------------
- fix disk detection bug (disk add, replace)

26.06b_rc
---------------------
- streamin tool name: cs_stream
- fix bug in create replication job
- check .info files (en and up to date)

26.06a_rc
---------------------
- Frontend Update (About > Frontend Update): reworked update flow --
  download/staging now fully separated from the actual stop/restart
  sequence, fixing instability where the update process could kill
  the web frontend mid-request
- Graceful backend/frontend shutdown on restart (replaces force-kill
  where possible -- faster restarts, no leftover stuck connections)
- Fixed a rare freeze in Frontend Update where a slow/unreachable
  download could stall the entire backend instead of timing out
- Frontend Update now also supports remote members (backend service
  only is restarted there, not the web frontend)

rc 10.2  (2026.06.28)
---------------------
- zstream replaces netcat for ZFS replication transport
  Encrypted AES-256-GCM, buffering (default 128 MB), rate limit, progress display
  https://github.com/guenther-alka/zstream  (BSD 2-Clause, Copyright napp-it.org)
- Per-job zstream settings: buffer size and rate limit (create + edit job)
- Transfer summary logged to /tmp/zstream.log after each replication
- Windows: job start via background() path (no Start-Process, works without desktop handle)
- task_norestart: fixed false "already running" block after cancel/error
- jobslib: transport always zstream, nc no longer supported
- server.pl: upsrv/callback path removed (curl upload to cs_connect.pl)
  Results now returned directly via encrypted socket (webserver.pl keeps socket open)
- admin.pl: Backend Update DIFFER false positive fixed (version extraction from 'on' response)
- socketlib.pl: remote socket timeout 30s -> 120s (large responses e.g. zfs get all ~5 MB)

rc 10.1  (2026.06)
-------------------
- Improved non-blocking behaviour of webserver.pl
- System > Backup/Restore: manual backup and restore of appliance settings
- Jobs > Appliance Backup: scheduled backup job with ZIP rotation (max 20)
- Disks > Enclosure Maps: visualisation of disk locations in backplane (set + printout)
- ChaCha20+HMAC-SHA256 encrypted socket between frontend and backend
- TOTP 2FA (RFC 6238, mobile authenticator app)
- Platform detection migrated from folder-based to $^O
- Hardcoded legacy paths replaced with /opt/csweb-gui/ across all files
- illumos/OmniOS distinction: fn="illumos" vs fn="solaris" for Oracle Solaris
- ZFS Servergroup: replaced sequential ping+socket checks with single socket call
  (worst-case 20-member check: ~300s -> ~20s)
- S3/RustFS: rc CLI replaces mc (AGPLv3 avoided), async bucket size loading
- L1/L2 two-level cache with stampede protection
- Session management: S_ID/S_LID dual-token, 50-minute rotation window
- Security: path traversal fix, shell injection fixes, XSS fix, brute-force protection