Joey Hess [Tue, 25 Jun 2024 21:26:26 +0000 (17:26 -0400)]
update
Joey Hess [Tue, 25 Jun 2024 21:20:49 +0000 (17:20 -0400)]
design for distributed clusters
Joey Hess [Tue, 25 Jun 2024 19:27:03 +0000 (15:27 -0400)]
thoughts on cycles
Rejected the idea of automatically instantiating remotes for proxies-of-proxies.
That needs cycle protection, while the current behavior, which happened
for free, is that running git-annex updateproxy on the proxy can be used
to configure it, but only for topologies that actually exist.
Joey Hess [Tue, 25 Jun 2024 18:52:47 +0000 (14:52 -0400)]
support annex.jobs for clusters
Joey Hess [Tue, 25 Jun 2024 18:10:06 +0000 (14:10 -0400)]
improve handling of cluster nodes disconnecting
Joey Hess [Tue, 25 Jun 2024 17:35:12 +0000 (13:35 -0400)]
gave up on upload fanout to cluster's proxy
The problem with that idea is that the cluster's proxy is necessarily a
remote, and necessarily one that we'll want to sync with, since the git
repository is stored there. So when its preferred content wants a file,
and the cluster does too, the file will get uploaded to it as well as to
the cluster. With fanout, the upload to the cluster will populate the
proxy as well, avoiding a second upload. But only if the file is sent to
the cluster first. If it's sent to the proxy first, there will be two
uploads.
Another, lesser problem is that a repository can proxy for more than one
cluster. So when does it make sense to drop content from the repository?
It could be done when dropping from one cluster, but what of the other
one?
This complication was not necessary anyway. Instead, if it's desirable
to have some content accessed from close to the proxy, one of the
cluster nodes can just be put on the same filesystem as it. That will be
just as fast as storing the content on the proxy.
Joey Hess [Tue, 25 Jun 2024 15:35:41 +0000 (11:35 -0400)]
honor preferred content settings of cluster nodes
Except when no nodes want a file, it has to be stored somewhere, so
store it on all. Which is not really desirable, but neither is having to
pick one.
ProtoAssociatedFile deserialization is rather broken, and this could
possibly affect preferred content expressions that match on filenames.
The inability to roundtrip whitespace like tabs and newlines through is
not a problem because preferred content expressions can't be written
that match on whitespace such as a tab. For example:
joey@darkstar:~/tmp/bench/z>git-annex wanted origin-node2 'exclude=*CTRL-VTab*'
wanted origin-node2
git-annex: Parse error: Parse failure: near "*"
But, the filtering of control characters could perhaps be a problem. I think
that filtering is now obsolete, git-annex has comprehensive filtering of
control characters when displaying filenames, that happens at a higher level.
However, I don't want to risk a security hole so am leaving in that filtering
in ProtoAssociatedFile deserialization for now.
Joey Hess [Tue, 25 Jun 2024 14:32:34 +0000 (10:32 -0400)]
PUT to cluster send to all nodes rather than none
If the location log says all nodes contain content, pass in all nodes,
rather than none.
The location log can be wrong. While it's good to avoid unncessessary
connections to nodes that already contain a key, it would be bad to
refuse to accept an upload at all when the location log is wrong.
Also, passing in no nodes leaves the proxy in an untenable state. It
can't proxy to no nodes. So it closes the connection. Passing in all
nodes means it has to do the work to connect to all of them, and see
that they say they already have the content, and then it can tell the
client that.
Joey Hess [Tue, 25 Jun 2024 14:06:28 +0000 (10:06 -0400)]
don't sync with cluster nodes by default
Avoid `git-annex sync --content` etc from operating on cluster nodes by default
since syncing with a cluster implicitly syncs with its nodes. This avoids a
lot of unncessary work when a cluster has a lot of nodes just in checking
if each node's preferred content is satisfied. And it avoids content
being sent to nodes individually, so instead syncing with clusters always
fanout uploads to nodes.
The downside is that there are situations where a cluster's preferred content
settings can be met, but those of its nodes are not. Or where a node does not
contain a key, but the cluster does, and there are not enough copies of the key
yet, so it would be desirable the send it there. I think that's an acceptable
tradeoff. These kind of situations are ones where the cluster itself should
probably be responsible for copying content to the node. Which it can do much
less expensively than a client can. Part of the balanced preferred content
design that I will be working on in a couple of months involves rebalancing
clusters, so I expect to revisit this.
The use of annex-sync config does allow running git-annex sync with a specific
node, or nodes, and it will sync with it. And it's also possible to set
annex-sync git configs to make it sync with a node by default. (Although that
will require setting up an explicit git remote for the node rather than relying
on the proxied remote.)
Logs.Cluster.Basic is needed because Remote.Git cannot import Logs.Cluster
due to a cycle. And the Annex.Startup load of clusters happens
too late for Remote.Git to use that. This does mean one redundant load
of the cluster log, though only when there is a proxy.
Joey Hess [Mon, 24 Jun 2024 14:13:13 +0000 (10:13 -0400)]
add annex-proxied
This makes git-annex sync and similar not treat proxied remotes as git
syncable remotes.
Also, display in git-annex info remote when the remote is proxied.
Joey Hess [Mon, 24 Jun 2024 13:40:57 +0000 (09:40 -0400)]
fix git-annex sync --content with proxied remotes
Loading the remote list a second time was removing all proxied remotes.
That happened because setting up the proxied remote added some config
fields to the in-memory git config, and on the second load, it saw those
configs and decided not to overwrite them with the proxy.
Now on the second load, that still happens. But now, the proxied
git configs are used to generate a remote same as if those configs were
all set. The reason that didn't happen before was twofold,
the gitremotes cache was not dropped, and the remote's url field was not
set correctly.
The problem with the remote's url field is that while it was marked as
proxy inherited, all other proxy inherited fields are annex- configs.
And the code to inherit didn't work for the url field.
Now it all works, but git-annex sync is left running git push/pull on
the proxied remote, which doesn't work. That still needs to be fixed.
Joey Hess [Mon, 24 Jun 2024 13:36:21 +0000 (09:36 -0400)]
drop gitremotes cache when config is changed
Joey Hess [Mon, 24 Jun 2024 13:35:24 +0000 (09:35 -0400)]
remove attempt to avoid git syncing with instantiate proxied remotes
It didn't work. Actually, sync was skipping those remotes due to a bug.
Joey Hess [Sun, 23 Jun 2024 20:38:01 +0000 (16:38 -0400)]
update
Joey Hess [Sun, 23 Jun 2024 20:22:39 +0000 (16:22 -0400)]
upload fanout resume seems free of fenceposts
Tested it with small chunk sizes (like 2) and resumes that were
eg 1 byte from the end of the file or beginning of file.
Also, git-annex testremote passes now against a cluster!
Joey Hess [Sun, 23 Jun 2024 20:01:58 +0000 (16:01 -0400)]
fix comparison
With this a PUT to two remotes that have different partial amounts
transferred works reliably. I'm not sure though that it doesn't have
fencepost errors.
Joey Hess [Sun, 23 Jun 2024 16:48:22 +0000 (12:48 -0400)]
update
Joey Hess [Sun, 23 Jun 2024 16:31:00 +0000 (12:31 -0400)]
update
Joey Hess [Sun, 23 Jun 2024 16:00:50 +0000 (12:00 -0400)]
always check numcopies when moving from cluster
When the destination does not start with a copy, the cluster has one or
more copies. If more, dropping would reduce the number of copies, so
numcopies must be checked.
Considered checking how many nodes of the cluster contain a copy. If
only 1 node does, it could allow a move without checking numcopies.
The problem with that, though, is that other nodes of the cluster could
have copies that we don't know about. And dropping from a cluster tries
to drop from all nodes, so will drop even from those. So any drop from a
cluster can remove more than 1 copy.
Joey Hess [Sun, 23 Jun 2024 14:09:11 +0000 (10:09 -0400)]
todo
Joey Hess [Sun, 23 Jun 2024 13:57:40 +0000 (09:57 -0400)]
don't use SUCCESS-PLUS unncessarily
When dropping from a proxied remote that is not a cluster,
SUCCESS-PLUS is not needed, so don't use it.
Joey Hess [Sun, 23 Jun 2024 13:53:33 +0000 (09:53 -0400)]
fix location log update for copy to 1-node cluster
Joey Hess [Sun, 23 Jun 2024 13:28:18 +0000 (09:28 -0400)]
dropping from clusters
Dropping from a cluster drops from every node of the cluster.
Including nodes that the cluster does not think have the content.
This is different from GET and CHECKPRESENT, which do trust the
cluster's location log. The difference is that removing from a cluster
should make 100% the content is gone from every node. So doing extra
work is ok. Compare with CHECKPRESENT where checking every node could
make it very expensive, and the worst that can happen in a false
negative is extra work being done.
Extended the P2P protocol with FAILURE-PLUS to handle the case where a
drop from one node succeeds, but a drop from another node fails. In that
case the entire cluster drop has failed.
Note that SUCCESS-PLUS is returned when dropping from a proxied remote
that is not a cluster, when the protocol version supports it. This is
because P2P.Proxy does not know when it's proxying for a single node
cluster vs for a remote that is not a cluster.
Joey Hess [Sun, 23 Jun 2024 12:21:11 +0000 (08:21 -0400)]
avoid storing SUCCESS-PLUS uuid when it is the remote uuid
This is slightly belt and suspenders, but nothing guarantees that the
peer avoids including its uuid in the SUCCESS-PLUS list as it's supposed
to. And while it probably doesn't matter if the location log is updated
redundantly, let's not find out.
Joey Hess [Sun, 23 Jun 2024 10:20:11 +0000 (06:20 -0400)]
avoid using cluster nodes in drop proof when dropping from cluster
This is obviously necessary in order for dropping from a cluster to be able to
drop from all nodes.
It also avoids violating numcopies when a cluster node is a special remote.
If it were used in the drop proof, nothing would prevent the cluster from
dropping from it.
Joey Hess [Sun, 23 Jun 2024 09:26:45 +0000 (05:26 -0400)]
update
Joey Hess [Thu, 20 Jun 2024 15:20:26 +0000 (11:20 -0400)]
Merge branch 'master' into proxy
Joey Hess [Thu, 20 Jun 2024 15:20:16 +0000 (11:20 -0400)]
merge from proxy branch
Joey Hess [Thu, 20 Jun 2024 15:03:30 +0000 (11:03 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Thu, 20 Jun 2024 15:03:19 +0000 (11:03 -0400)]
add my distribits talk
Joey Hess [Thu, 20 Jun 2024 14:57:43 +0000 (10:57 -0400)]
clusters documentation
Joey Hess [Thu, 20 Jun 2024 14:15:24 +0000 (10:15 -0400)]
wording
Joey Hess [Thu, 20 Jun 2024 14:04:26 +0000 (10:04 -0400)]
working PUT fanout to multiple remotes for clusters
Still need to check for fencepost errors on resume when
different nodes have different amounts of data.
joris [Thu, 20 Jun 2024 09:58:05 +0000 (09:58 +0000)]
Added a comment
Joey Hess [Wed, 19 Jun 2024 10:40:19 +0000 (06:40 -0400)]
more on proxying special remotes
Joey Hess [Wed, 19 Jun 2024 10:15:03 +0000 (06:15 -0400)]
towards a design for proxying to special remotes
Joey Hess [Wed, 19 Jun 2024 01:31:32 +0000 (21:31 -0400)]
be quiet when reading cluster and proxy information at startup
I had a transfer of 3 files fail like this:
git-annex: transferrer protocol error: "(recording state in git...)"
The remote had stalldetection enabled, although I didn't see it stall.
So git-annex transferrer would have been started up. I guess that
one of these new git-annex branch reads, that happens early, caused
that message due to perhaps an uncommitted git-annex branch change.
Since the transferrer speaks a protocol over stdout, it needs to be
prevented from outputting other messages to stdout. Interestingly,
startupAnnex is run after prepRunCommand, so if a command requests quiet
output it would already be quiet. But the transferrer does not, instead
it calls Annex.setOutput SerializedOutput in its start action.
Joey Hess [Tue, 18 Jun 2024 21:02:23 +0000 (17:02 -0400)]
allow proxying to remotes that are nodes of clusters
fixes reversion in
ca08f3fcc2ca6d203fa1a323986f7f86bd342e3d
Joey Hess [Tue, 18 Jun 2024 16:07:01 +0000 (12:07 -0400)]
P2P protocol version 2, adding SUCCESS-PLUS and ALREADY-HAVE-PLUS
Client side support for SUCCESS-PLUS and ALREADY-HAVE-PLUS
is complete, when a PUT stores to additional repositories
than the expected on, the location log is updated with the
additional UUIDs that contain the content.
Started implementing PUT fanout to multiple remotes for clusters.
It is untested, and I fear fencepost errors in the relative
offset calculations. And it is missing proxying for the protocol
after DATA.
Joey Hess [Tue, 18 Jun 2024 15:43:10 +0000 (11:43 -0400)]
only proxy to a remote when remote.name.annex-proxy is set
Avoids someone writing to proxy.log and gaining access to remotes
of someone else's repository that they were not intended to be able
to proxy to.
Joey Hess [Tue, 18 Jun 2024 15:37:38 +0000 (11:37 -0400)]
only use a remote as a node when git configuration is set
Avoids someone writing to cluster.log and nominating remotes
of someone else's repository as a cluster.
Joey Hess [Tue, 18 Jun 2024 15:10:48 +0000 (11:10 -0400)]
checkpresent support for clusters
This assumes that the proxy for a cluster has up-to-date location
logs. If it didn't, it might proxy the checkpresent to a node that no
longer has the content, while some other node still does, and so
it would incorrectly appear that the cluster no longer contains the
content.
Since cluster UUIDs are not stored to location logs,
git-annex fsck --fast when claiming to fix a location log when
that occurred would not cause any problems. And presumably the location
tracking would later get sorted out.
At least usually, changes to the content of nodes goes via the proxy,
and it will update its location logs, so they will be accurate. However,
if there were multiple proxies to the same cluster, or nodes were
accessed directly (or via proxy to the node and not the cluster),
the proxy's location log could certainly be wrong.
(The location log access for GET has the same issues.)
Joey Hess [Tue, 18 Jun 2024 15:01:10 +0000 (11:01 -0400)]
initial, working support for getting from clusters
Currently tends to put all the load on a single node, which will need to
be improved.
Joey Hess [Tue, 18 Jun 2024 14:51:37 +0000 (10:51 -0400)]
factor out Annex.Proxy
Joey Hess [Tue, 18 Jun 2024 14:36:04 +0000 (10:36 -0400)]
refactor cluster code into own module
Joey Hess [Tue, 18 Jun 2024 14:08:15 +0000 (10:08 -0400)]
update
yarikoptic [Tue, 18 Jun 2024 13:57:23 +0000 (13:57 +0000)]
original report / question
Joey Hess [Mon, 17 Jun 2024 23:19:15 +0000 (19:19 -0400)]
ProxySelector data type
Joey Hess [Mon, 17 Jun 2024 19:51:10 +0000 (15:51 -0400)]
preparing for cluster node selection
Support selecting what remote to proxy for each top-level P2P protocol
message.
This only needs to be extended now to support fanout to multiple
nodes for PUT and REMOVE, and with a remote that fails for
LOCKCONTENT and UNLOCKCONTENT.
But a good first step would be to implement CHECKPRESENT and GET for
clusters. Both should select a node that actually does have the content.
That will allow a cluster to work for GET even when location tracking is
out of date.
Joey Hess [Mon, 17 Jun 2024 19:00:11 +0000 (15:00 -0400)]
started on git-annex-shell cluster support
Works down to P2P protocol.
The question now is, how to handle protocol version negotiation for
clusters? Connecting to each node to find their protocol versions and
using the lowest would be too expensive with a lot of nodes. So it seems
that the cluster needs to pick its own protocol version to use with the
client.
Then it can either negotiate that same version with the nodes when
it comes time to use them, or it can translate between multiple protocol
versions. That seems complicated. Thinking it would be ok to refuse to
use a node if it is not able to negotiate the same protocol version with
it as with the client. That will mean that sometimes need nodes to be
upgraded when upgrading the cluster's proxy. But protocol versions
rarely change.
Joey Hess [Mon, 17 Jun 2024 18:14:08 +0000 (14:14 -0400)]
work toward supporting proxying to multiple remotes at once
For eg, upload fanout.
Delay connecting to a remote until it's needed. When there are many
proxied remotes, it would not do for the proxy to connect to each of
them on startup; that could take a long time.
Joey Hess [Mon, 17 Jun 2024 17:04:40 +0000 (13:04 -0400)]
more specific type
Joey Hess [Mon, 17 Jun 2024 16:44:08 +0000 (12:44 -0400)]
improve types
Joey Hess [Mon, 17 Jun 2024 13:31:44 +0000 (09:31 -0400)]
update
Joey Hess [Mon, 17 Jun 2024 13:29:34 +0000 (09:29 -0400)]
Merge branch 'master' into proxy
Joey Hess [Mon, 17 Jun 2024 13:26:03 +0000 (09:26 -0400)]
tab complete special remotes
An oversight..
And with the work in progress proxy and cluster, there
can be additional remotes that are not listed in .git/config, but are
available. Making those more discoverable is another big benefit of
this.
Joey Hess [Sun, 16 Jun 2024 19:07:48 +0000 (15:07 -0400)]
don't count clusters as copies, continued
Handled limitCopies, as well as everything using fromNumCopies and
fromMinCopies.
This should be everything, probably.
Note that, git-annex info displays a count of repositories, which still
includes cluster. I think that's ok. It would be possible to filter out
clusters there, but to the user they're pretty much just another
repository. The numcopies displayed by eg `git-annex info .` does not
include clusters.
Joey Hess [Sun, 16 Jun 2024 18:35:07 +0000 (14:35 -0400)]
remove dead nodes when loading the cluster log
This is to avoid inserting a cluster uuid into the location log when
only dead nodes in the cluster contain the content of a key.
One reason why this is necessary is Remote.keyLocations, which excludes
dead repositories from the list. But there are probably many more.
Implementing this was challenging, because Logs.Location importing
Logs.Cluster which imports Logs.Trust which imports Remote.List resulted
in an import cycle through several other modules.
Resorted to making Logs.Location not import Logs.Cluster, and instead
it assumes that Annex.clusters gets populated when necessary before it's
called.
That's done in Annex.Startup, which is run by the git-annex command
(but not other commands) at early startup in initialized repos. Or,
is run after initialization.
Note that is Remote.Git, it is unable to import Annex.Startup, because
Remote.Git importing Logs.Cluster leads the the same import cycle.
So ensureInitialized is not passed annexStartup in there.
Other commands, like git-annex-shell currently don't run annexStartup
either.
So there are cases where Logs.Location will not see clusters. So it won't add
any cluster UUIDs when loading the log. That's ok, the only reason to do
that is to make display of where objects are located include clusters,
and to make commands like git-annex get --from treat keys as being located
in a cluster. git-annex-shell certainly does not do anything like that,
and I'm pretty sure Remote.Git (and callers to Remote.Git.onLocalRepo)
don't either.
Joey Hess [Sun, 16 Jun 2024 15:34:35 +0000 (11:34 -0400)]
don't count clusters as copies
Since the cluster UUID is inserted into the location log when the
location log lists a node as containing content.
Also avoid trying to lock content on cluster remotes. The cluster nodes
are also proxied, so that content can be locked on individual nodes, and
locking content on a cluster as a whole probably won't be implemented.
And made git-annex whereis use numcopies machinery for displaying its
count, so it won't count cluster UUIDs redundantly to nodes.
Other commands, like git-annex info that also display numcopies
information already used the numcopies machinery.
There is more to be done, fromNumCopies is sometimes used to get a
number that is compared with a list of UUIDs. And limitCopies doesn't
use numcopies machinery.
Added a comment
Added a comment: Grafting? a special remote for tuned migration
Joey Hess [Fri, 14 Jun 2024 22:06:28 +0000 (18:06 -0400)]
insert cluster UUIDs when loading location logs, and omit when saving
Inline isClusterUUID for speed.
Joey Hess [Fri, 14 Jun 2024 21:27:34 +0000 (17:27 -0400)]
remove Logs.Presence imports
When imported along with Logs.Location, it can be an unused import and
it won't warn, due to reexports. The point if this is really to show
that Logs.Presence is not widely used, outside Logs/
Joey Hess [Fri, 14 Jun 2024 21:13:23 +0000 (17:13 -0400)]
broke out initcluster
One benefit of this is that a typo in annex-cluster-node config won't
init a new cluster.
Also it gets the cluster description set and is consistent with
initremote.
Joey Hess [Fri, 14 Jun 2024 20:37:17 +0000 (16:37 -0400)]
fogot to add
Joey Hess [Fri, 14 Jun 2024 19:23:43 +0000 (15:23 -0400)]
update todo list for this month
whew that's gonna be a lot
Joey Hess [Fri, 14 Jun 2024 19:03:20 +0000 (15:03 -0400)]
add clusters to proxy log
Note that it's not defined what will happen if a cluster has the same
name as a remote that has proxying enabled.
Joey Hess [Fri, 14 Jun 2024 18:21:50 +0000 (14:21 -0400)]
add git-annex updatecluster command
Seems to work fine, making the right changes to the git-annex branch.
Joey Hess [Fri, 14 Jun 2024 15:42:32 +0000 (11:42 -0400)]
add git configs for clusters
Joey Hess [Fri, 14 Jun 2024 15:16:01 +0000 (11:16 -0400)]
cache getClusters in Annex state
Joey Hess [Fri, 14 Jun 2024 15:07:31 +0000 (11:07 -0400)]
make cluster UUIDs distinguishable from any other repository UUID
A cluster UUID is a version 8 UUID, with first octets 'a' and 'c'.
The rest of the content will be random.
This avoids a class of attack where the UUID of a repository is used as
the UUID of a cluster, which will prevent git-annex from updating
location logs for that repository. I don't know why someone would want
to do that, but let's prevent it.
Also, isClusterUUID make it easy to filter out cluster UUIDs when
writing the location logs.
Joey Hess [Thu, 13 Jun 2024 23:08:04 +0000 (19:08 -0400)]
update
Joey Hess [Thu, 13 Jun 2024 21:56:53 +0000 (17:56 -0400)]
unique uuid namespace for clusters
Joey Hess [Thu, 13 Jun 2024 20:00:58 +0000 (16:00 -0400)]
implement cluster.log
Not used yet. (Or tested.)
I did consider making the log start with the uuid of the node, followed
by the cluster uuid (or uuids). That would perhaps mean a smaller write
to the git-annex branch when adding a node, but overall the log file
would be larger, and it will be read and cached near to startup on most
git-annex runs.
Joey Hess [Thu, 13 Jun 2024 18:30:32 +0000 (14:30 -0400)]
comment
Joey Hess [Thu, 13 Jun 2024 17:40:04 +0000 (13:40 -0400)]
comment
Joey Hess [Thu, 13 Jun 2024 17:03:38 +0000 (13:03 -0400)]
partial reproducer
Joey Hess [Thu, 13 Jun 2024 15:44:39 +0000 (11:44 -0400)]
update
Joey Hess [Thu, 13 Jun 2024 14:48:31 +0000 (10:48 -0400)]
more cluster thoughts
Joey Hess [Thu, 13 Jun 2024 14:34:19 +0000 (10:34 -0400)]
avoided the strangeness of the cluster's proxy location tracking being wrong
Joey Hess [Thu, 13 Jun 2024 10:49:36 +0000 (06:49 -0400)]
update
Joey Hess [Thu, 13 Jun 2024 10:44:22 +0000 (06:44 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Thu, 13 Jun 2024 10:43:59 +0000 (06:43 -0400)]
copied over some changes from proxy branch
Joey Hess [Thu, 13 Jun 2024 10:41:42 +0000 (06:41 -0400)]
more thoughts on clusters
Joey Hess [Wed, 12 Jun 2024 21:30:55 +0000 (17:30 -0400)]
more thoughts on clusters
Joey Hess [Wed, 12 Jun 2024 19:21:23 +0000 (15:21 -0400)]
update
Joey Hess [Wed, 12 Jun 2024 19:10:03 +0000 (15:10 -0400)]
avoid git syncing with instantiate proxied remotes
These remotes have no url configured, so git pull and push will fail.
git-annex sync --content etc can still sync with them otherwise.
Also, avoid git syncing twice with the same url. This is for cases where
a proxied remote has been manually configured and so does have a url.
Or perhaps proxied remotes will get configured like that automatically
later.
Joey Hess [Wed, 12 Jun 2024 18:45:39 +0000 (14:45 -0400)]
designing clusters
Joey Hess [Wed, 12 Jun 2024 17:52:17 +0000 (13:52 -0400)]
on cycles
Joey Hess [Wed, 12 Jun 2024 17:24:31 +0000 (13:24 -0400)]
dash is legal in git remote names
Joey Hess [Wed, 12 Jun 2024 17:24:25 +0000 (13:24 -0400)]
whitespace
Joey Hess [Wed, 12 Jun 2024 17:06:14 +0000 (13:06 -0400)]
add missing spaces
Joey Hess [Wed, 12 Jun 2024 16:37:14 +0000 (12:37 -0400)]
update
Joey Hess [Wed, 12 Jun 2024 15:55:18 +0000 (11:55 -0400)]
updates
Joey Hess [Wed, 12 Jun 2024 15:53:28 +0000 (11:53 -0400)]
inherit remote.name.annex-bare
Since a proxied remote uses the proxy's git repo, this makes sense.
Although I don't think this config is ever used when accessing a remote
via git-annex-shell.
m.risse@77eac2c22d673d5f10305c0bade738ad74055f92 [Wed, 12 Jun 2024 15:42:42 +0000 (15:42 +0000)]
Joey Hess [Wed, 12 Jun 2024 15:37:14 +0000 (11:37 -0400)]
proxy updates location tracking information
This does mean a redundant write to the git-annex branch. But,
it means that two clients can be using the same proxy, and after
one sends a file to a proxied remote, the other only has to pull from
the proxy to learn about that. It does not need to pull from every
remote behind the proxy (which it couldn't do anyway as git repo
access is not currently proxied).
Anyway, the overhead of this in git-annex branch writes is no worse
than eg, sending a file to a repository where git-annex assistant
is running, which then sends the file on to a remote, and updates
the git-annex branch then. Indeed, when the assistant also drops
the local copy, that results in more writes to the git-annex branch.
Joey Hess [Wed, 12 Jun 2024 14:40:51 +0000 (10:40 -0400)]
finish P2P protocol proxying
CONNECT is not supported by git-annex-shell p2pstdio, but for proxying
to tor-annex remotes, it will be supported, and will make a git pull/push
to a proxied remote work the same with that as it does over ssh,
eg it accesses the proxy's git repo not the proxied remote's git repo.
The p2p protocol docs say that NOTIFYCHANGES is not always supported,
and it looked annoying to implement it for this, and it also seems
pretty useless, so make it be a protocol error. git-annex remotedaemon
will already be getting change notifications from the proxy's git repo,
so there's no need to get additional redundant change notifications for
proxied remotes that would be for changes to the same git repo.
Joey Hess [Wed, 12 Jun 2024 14:16:04 +0000 (10:16 -0400)]
a local git remote cannot proxy
Prevent listProxied from listing anything when the proxy remote's
url is a local directory. Proxying does not work in that situation,
because the proxied remotes have the same url, and so git-annex-shell
is not run when accessing them, instead the proxy remote is accessed
directly.
I don't think there is any good way to support this. Even if the instantiated
git repos for the proxied remotes somehow used an url that caused it to use
git-annex-shell to access them, planned features like `git-annex copy --to
proxy` accepting a key and sending it on to nodes behind the proxy would not
work, since git-annex-shell is not used to access the proxy.
So it would need to use something to access the proxy that causes
git-annex-shell to be run and speaks P2P protocol over it. And we have that.
It's a ssh connection to localhost. Of course, it would be possible to
take ssh out of that mix, and swap in something that does not have
encryption overhead and authentication complications, but otherwise
behaves the same as ssh. And if the user wants to do that, GIT_SSH
does exist.