Re: Can unison better help with renamed hosts?
"'Mario Emmenlauer' via unison-users" <[email protected]> Wed, 24 Apr 2024 14:39:52 +0200
| Newsgroups | gmane.network.unison.general |
|---|---|
| Message-ID | <[email protected]> |
On 24.04.24 14:31, 'Mario Emmenlauer' via unison-users wrote: > On 24.04.24 14:25, Tõivo Leedjärv wrote: >> On Wed, 24 Apr 2024 at 14:20, 'Mario Emmenlauer' via unison-users >> <[email protected]> wrote: >>> >>> All I would need to follow this up is >>> the printed value, to help me understand the root cause. >>> >>> Or am I on the wrong track here? >> >> I think you are :) >> When you get the warning, the names are stated right there, e.g. >> >> Warning: No archive files were found for these roots, whose canonical names are: >> (the names are here!) > > I see, and for what its worth, I do feel a bit silly now! :-) :-) > > But then again, the printed host name _was_ the one that is in the > archive file! It was just "bool" and nothing else. And that is what > hostname consistently reports. Ok, I may have been a bit blind here. Which host names are considered in computing the archive checksum? Only the remote, or both local and remote? I was naively assuming that the local machine is not expected to change its host name, and even if it would, that should not matter for unison. But my findings now would make perfect sense if the local hostname is considered! This can very likely change (from DHCP) when the laptop leaves the LAN. To support this finding, I made the following observations. I sync two computers against our server "bool". One machine is "decuman", inside the LAN. And one is "eilian", outside the LAN. They show: #eilian> head -2 .unison/are67e2ab3276b3050080e50b64dd99522 Unison archive format 23 Archive for root //eilian// synchronizing roots //bool//mnt/unisonsync, //eilian// #decuman> head -2 .unison/ar19e593e1cc370e9f45bb2c08b9a1fe6d Unison archive format 23 Archive for root //decuman// synchronizing roots //bool//mnt/unisonsync, //decuman// As you can see, the archive checksums are different on the two hosts, albeit they should see "bool" with the same hostname (if queried on bool). The main difference is their own hostname. So, is it correct that the local machine hostname is involved? That would be slightly unfortunate for my laptop use case, because it may change the domain name dynamically with the DHCP server... All the best, Mario To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].