When the server stops responding

Networks drop and iSCSI services restart. BeamDisk is built to ride that out, the way the iSCSI initiators in Windows and Linux do: it holds writes, so your apps wait instead of failing.

What happens

Writes wait. When the server stops answering, the disk’s state turns to Reconnecting and BeamDisk holds its writes for up to 60 seconds. Apps wait as they would for a slow disk. The menu-bar item shows an amber dot, and the disk says what’s going on, with since when: The server stopped responding. Holding writes — 12 of 60 s; nothing is lost yet. It may also say how much is waiting: 48 MB of recent writes are waiting for the server.
BeamDisk reconnects by itself, logs in again (with CHAP, if you use it) and carries on where it stopped. It tries again after a second, then a little less often after each failed try, never more than 10 seconds apart. When the disk is back, you get a notification: Projects reconnected.
If the outage outlasts the hold time, apps start to see errors on that disk. Its state turns to Apps see errors, the dot turns red, and BeamDisk keeps trying to reconnect until you disconnect the disk.

In testing, iSCSI service restarts, network outages and forced disconnects lost no data that the server had acknowledged.

While it’s reconnecting

The Overview puts the disk at the top, with a bar filling up over the hold time and a Disconnect button. The disk’s page says the same, with Eject. In the quick menu, its row counts: Holding writes · 12 of 60 s. You don’t need to do anything: most outages end within the hold time.

Disconnect or eject a reconnecting disk only if you don’t expect the server back, and have nothing unsaved on it.

Change the hold time

In Settings → Connection, Hold writes for sets how long writes wait, from 15 to 300 seconds in steps of 5. A longer time rides out longer outages, such as a server restart; a shorter one tells apps sooner that the disk is gone. The change applies at once, to every disk. A disk’s page repeats the time in its Connection card: If the server stops responding, writes are held for up to 60 s before apps see errors.

If a disk says Remount needed

After an outage that outlasted the hold time, a disk can come back as Remount needed: The server was unreachable for longer than the hold time. Remount to be sure the volume is consistent. Writes made during the outage that hadn’t reached the server may be lost, as with a drive that was unplugged. Click Remount to eject the disk and attach it again.

If the server lost power

A server that loses power or restarts unexpectedly can lose writes it had accepted but not yet saved. On disks where it’s on, BeamDisk’s power-loss protection keeps your most recent writes safe through that, as long as your Mac stays on. It’s on by default for TrueNAS.