Commit graph

126 commits

Author SHA1 Message Date
ba6b56ae68
Fix some new clippy lints 2022-03-14 12:27:49 +01:00
2377a92f6b
Add wrapper over sled tree to count items (used for big queues) 2022-03-14 10:54:25 +01:00
203e8d2c34
Bump version to 0.7 because of incompatible Netapp 2022-03-14 10:54:24 +01:00
f869ca625d
Add spans to table calls, change span names in RPC 2022-03-14 10:54:12 +01:00
0cc31ee169
add missing netapp telemetry feature 2022-03-14 10:54:11 +01:00
dc8d0496cc
Refactoring: rename config files, make modifications less invasive 2022-03-14 10:53:51 +01:00
2a5609b292
Add metrics to API endpoint 2022-03-14 10:53:36 +01:00
818daa5c78
Refactor how durations are measured 2022-03-14 10:53:35 +01:00
bb04d94fa9
Update to Netapp 0.4 which supports distributed tracing 2022-03-14 10:52:30 +01:00
8c2fb0c066
Add tracing integration with opentelemetry 2022-03-14 10:52:13 +01:00
2cab84b1fe
Add many metrics in table/ and rpc/ 2022-03-14 10:51:50 +01:00
9d44127245
add support for kubernetes service discovery
This commit adds support to discover garage instances running in
kubernetes.

Once enabled by setting `kubernetes_namespace` and
`kubernetes_service_name` garage will create a Custom Resources
`garagenodes.deuxfleurs.fr` with nodes public key as the resource name.
and IP and Port information as spec in the namespace configured by
`kubernetes_namespace`.

For discovering nodes the resources are filtered with the optionally set
`kubernetes_service_name` which sets a label
`garage.deuxfleurs.fr/service` on the resources.

This allows to separate multiple garage deployments in a single
namespace.

the `kubernetes_skip_crd` variable allows to disable the creation of the
CRD by garage itself. The user must deploy this manually.
2022-03-12 13:05:52 +01:00
beeef4758e
Some movement of helper code and refactoring of error handling 2022-01-04 12:52:46 +01:00
5b1117e582
New model for buckets 2022-01-04 12:45:46 +01:00
c94406f428
Improve how node roles are assigned in Garage
- change the terminology: the network configuration becomes the role
  table, the configuration of a nodes becomes a node's role
- the modification of the role table takes place in two steps: first,
  changes are staged in a CRDT data structure. Then, once the user is
  happy with the changes, they can commit them all at once (or revert
  them).
- update documentation
- fix tests
- implement smarter partition assignation algorithm

This patch breaks the format of the network configuration: when
migrating, the cluster will be in a state where no roles are assigned.
All roles must be re-assigned and commited at once. This migration
should not pose an issue.
2021-11-16 16:05:53 +01:00
e8811f7c9d
Request strategy: don't launch all 3 requests if not needed 2021-11-04 16:19:27 +01:00
6f13d083ab
Add semaphore to limit RAM used by buffered outgoing requests 2021-11-03 18:02:57 +01:00
8c4f418fe8
Fix peer list persistence: do not forget previous peers 2021-11-03 17:34:44 +01:00
43e13a501d
Use published netapp crate instead of git repo 2021-10-26 10:36:57 +02:00
ada7899b24
Fix clippy lints (fix #121) 2021-10-26 10:20:05 +02:00
de4276202a
Improve CLI, adapt tests, update documentation 2021-10-25 14:21:48 +02:00
1b450c4b49
Improvements to CLI and various fixes for netapp version
Discovery via consul, persist peer list to file
2021-10-22 16:55:24 +02:00
4067797d01
First port of Garage to Netapp 2021-10-22 15:55:18 +02:00
fa394dcd27
Support pkcs8 private keys (allowing for ed25519 to be used for rpc) 2021-07-06 11:16:01 +02:00
30a7dee920 exit when inconsistent level of replication is detected (#92)
fix #88

Authored-by: Trinity Pointard <trinity.pointard@gmail.com>
Reviewed-on: Deuxfleurs/garage#92
Co-authored-by: trinity-1686a <trinity.pointard@gmail.com>
Co-committed-by: trinity-1686a <trinity.pointard@gmail.com>
2021-06-02 13:30:39 +02:00
289521886b make most changes suggested during install-party 2021-05-29 21:37:49 +02:00
b9127dd6f8
Prepare for v0.3.0 and add migration path from v0.2.1.x 2021-05-28 15:29:58 +02:00
ddb2b29bfd
Rename datacenters into zones (doc not yet updated) 2021-05-28 14:07:36 +02:00
b490ebc7f6
Many improvements on ring/replication and its configuration:
- Explicit "replication_mode" configuration parameters that takes
  either "none", "2" or "3" as values, instead of letting user configure
  replication factor themselves. These are presets whose corresponding
  replication/quorum values can be found in replication/mode.rs

- Explicit support for single-node and two-node deployments
  (number of nodes must be at least "replication_mode", with "none"
  we can have only one node)

- Ring is now stored much more compactly with 256*8 + n*32 bytes,
  instead of 256*32 bytes

- Support for gateway-only nodes that do not store data
  (these nodes still need a metadata_directory to store the list
  of bucket and keys since those are stored on all nodes; it also
  technically needs a data_directory to start but it will stay
  empty unless we have bugs)
2021-05-28 14:07:36 +02:00
e4b9e4e24d
rename types to CamelCase 2021-05-03 22:15:09 +02:00
f05bb111c2
fix clippy warnings on util and rpc 2021-05-03 22:11:41 +02:00
2812a027ea
change some more comments and revert changes on TableSchema 2021-04-27 16:49:07 +02:00
74373aebcf
make most requested changes 2021-04-27 16:47:08 +02:00
16300bbd89
remove useless comment 2021-04-27 16:44:01 +02:00
f871689571
run cargo fmt on util and make missing doc warning 2021-04-27 16:37:10 +02:00
8e0524ae15
document rpc crate 2021-04-27 16:37:10 +02:00
6b2b400292
small simplify 2021-04-27 16:37:09 +02:00
8c33d565d6
Merge discovery loop with consul 2021-04-27 16:37:09 +02:00
948e44a3f6
cargo fmt 2021-04-27 16:37:09 +02:00
3e2e38c830
Print stats 2021-04-27 16:37:09 +02:00
2e53e31cdd
Cargo fmt 2021-04-27 16:37:09 +02:00
64b91c2645
Keep old data 2021-04-27 16:37:09 +02:00
e16077f40a
Persist directly and not in background 2021-04-27 16:37:09 +02:00
9ced9f78dc
Improve bootstraping: do it regularly; persist peer list 2021-04-27 16:37:08 +02:00
f859d15062 update to v0.2.1 2021-03-19 13:39:18 +01:00
4c26a0b9c1 Update Cargo.toml files with AGPL license info 2021-03-18 21:59:17 +01:00
dead945c8f Prepare for release 0.2 2021-03-18 19:33:15 +01:00
f4346cc5f4 Update dependencies 2021-03-16 15:58:40 +01:00
2a41b82384 Simpler Merkle & sync 2021-03-16 12:18:03 +01:00
1d9961e411 Simplify replication logic 2021-03-16 11:14:27 +01:00
6a8439fd13 Some improvements in background worker but we terminate late 2021-03-15 23:14:12 +01:00
0cd5b2ae19 WIP migrate to tokio 1 2021-03-15 22:36:41 +01:00
4d4117f2b4 Refactor block resync loop; make workers infaillible 2021-03-15 20:09:44 +01:00
537f652fec Tiny things 2021-03-15 18:40:27 +01:00
3bf2df622a Time and metadata improvements 2021-03-15 16:21:41 +01:00
c475471e7a Implement table gc, currently for block_ref and version only 2021-03-12 19:57:37 +01:00
046b649bcc (not well tested) use merkle tree for sync 2021-03-11 18:28:27 +01:00
8d63738cb0 Checkpoint: add merkle tree in data table 2021-03-11 13:47:21 +01:00
3214dd52dd Very minor changes 2021-03-10 21:50:09 +01:00
6a3dcf3974 Rename n_tokens into capacity 2021-03-10 14:52:03 +01:00
7cda917b6b update condition 2021-03-05 17:08:03 +01:00
d7e005251d Not fully tested: new multi-dc MagLev 2021-03-05 16:22:29 +01:00
20e6e9fa20 Update sled & try to debug deadlock (but its in sled...) 2021-02-23 21:27:28 +01:00
40763fd749 Cargo fmt 2021-02-23 18:46:25 +01:00
6e6f7e8555 Replace some checksums where it makes sense 2021-02-23 18:14:37 +01:00
b1b640ae8b rename hash() to sha256sum(), we might want to change it at some places 2021-02-21 15:24:30 +01:00
80892df8cc Some refactoring 2021-02-21 13:11:10 +01:00
1d1d497e2b Bump everything to 0.1.1 2021-01-15 17:54:48 +01:00
8956db2a81 Make less things public 2020-12-12 17:58:19 +01:00
a50fa70d45 Refactor error management in API part 2020-11-08 15:05:28 +01:00
3b0b11085e Add versions to dependencies 2020-07-07 14:18:47 +02:00
cc65cdc0fe Add license, description and repository to .toml files 2020-07-07 14:14:58 +02:00
fbe8fe81f2 Add automatic peer discovery from Consul 2020-06-30 18:33:14 +02:00
16fbb32fd3 Rate limit requests a bit more seriously
droping the slot later (after reading the request response)
means that we aren't freeing our quota slot,
so the maximum number of simultaneous requests now also counts the
response reading phase

TODO next: quotas per rpc destination node, or maybe per datacenter (?)
2020-05-01 19:18:54 +00:00
d8f5e643bc Split code for modular compilation 2020-04-24 10:10:01 +00:00
c9c6b0dbd4 Reorganize code 2020-04-23 17:05:46 +00:00