Open
Conversation
Solution: Modify hatch configuration to have the environmnent properly set up using the virtual environment builtin module https://hatch.pypa.io/1.3/plugins/environment/virtual/
It's a problem surfaced by another The visible problem was that the new exection test were hanging, inside the runtime, during the import at the line from aleph.sdk.chains.remote import RemoteAccount after some more investigative work, it was pin pointed to an inner import of eth_utils module (specifically eth_utils.network ) Second problem that made the first visible: in the runtime the pre-compiled bytecode, created during runtime creation in create_disk_image.sh was not used, which made the import of module slower. This surfaced the first problem. The cause of that second problem was that the init1.py code which run the user caude was not launched with the same optimization level as the pre-compiled bytecode and thus recompiled everything. (this is specified in the init1.py #! sheebang on the first line) Solution: Compile the bytecode with the same optimisation level (-o 2 ) as during run We haven't found out yet why the eth_utils.network import hang when it is not precompiler. But this fix the test hanging issue
…asyncio' Fix: async_sessionmaker was introduced in sqlachemy 2.0, ensure we have at least this version otherwhise it was using a older system package
A frozen copy of the requirements.txt extracted from different systems was present in the repository but not used nor maintained.
|
Failed to retrieve llama text: POST 504: 504 Gateway Time-outThe server didn't respond in time. |
hoh
force-pushed
the
hoh-update-vm-connector
branch
from
cc6f432 to
6b1781f
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #594 +/- ##
=======================================
Coverage 64.50% 64.50%
=======================================
Files 78 78
Lines 7088 7088
Branches 598 598
=======================================
Hits 4572 4572
Misses 2314 2314
Partials 202 202 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Internet connectivity checks by the diagnostic VM relied on a single URL. If that endpoint was down, the internet connectivity of the system was assumed to be down. Solution: Check connectivity to multiple endpoints in parallel.
'aleph program' now need an 'update' argument. Solution: Update makefile and documentation
Problem: could not start Instances from command line Problem happened when launching with --run-fake-instance Solution: Adapt to new VMPool API that take a loop Also fix benchmarks function
Fix: Solve last CORS errors raised cause by duplication of headers returned.
We published multiple changes to the diagnostic VM recently but none of these was released. This provides a new diagnostic VM, based on a new runtime [1], with fixes: - Reading messages with the newer SDK - Better handling of IPv6 detection errors - Two different tests for signing messages (local and remote) - aleph-message version was not specified - fetching a single message was not tested
Point to it in the documentation and removed duplicated information here
Solution: Start by adding some simple tests We don't test the full allocation and deallocation here. just auth
Co-authored-by: nesitor <amolinsdiaz@yahoo.es>
When executed using `bash`, the `create_disk_image` was interrupted by a Python REPL due to the `python -OO` command being surrounded by backquotes.
Problem: Running system testing on DigitalOcean for every push consumed a lot of resources and failed frequently. We now start to have integration testing using `pytest`, which provides a better confidence that things actually work. Solution: Only test on DO open Pull Requests and not every push. In the future, consider only running when merged on `main`.
This bumps their versions to - Kubo 0.23.0 -> 0.33.1 The list of changes regarding Kubo too large to be mentioned here, but I mostly expect performance improvements as the main API has not changed much.
Due to an error reading the denylist. The error was: ``` Error: constructing the node (see log for full detail): error walking /home/ipfs/.config/ipfs/denylists: lstat /home/ipfs/.config/ipfs/denylists: permission denied ``` The [documentation](https://specs.ipfs.tech/compact-denylist-format/) mentions that: > Implementations SHOULD look in /etc/ipfs/denylists/ and > $XDG_CONFIG_HOME/ipfs/denylists/ (default: ~/.config/ipfs/denylists) > for denylist files. I am not sure why this only failed on Ubuntu 22.04 and not Debian 12 or Ubuntu 24.04. My first assumption would be a difference in Systemd.
hoh
force-pushed
the
hoh-update-vm-connector
branch
from
493337e to
df70293
Compare
hoh
force-pushed
the
hoh-update-vm-connector
branch
from
df70293 to
03b780b
Compare
Needed for some crn
* Make CI a bit more resilient Do not run export log if cancelled or not setup Attempt to always do the proper clean up Print more debug information in case the droplet ipv4 cannot be parsed * CI: Add timeouts for runtime workflow * CI: Document the runtime workflow * CI: Increase timeout to bring up droplet * CI: pytest. Prevent "Ouput modules" step failure The dep were not always installed depending on where the failure was Since this a debug step we don't want it to fail Do not run it if the workflow was cancelled * CI: Merge Package workflow and Droplet workflow This allow to reuse the packages build in the previous workflow for the droplet test, reducing the number of package build from 12 to 3, in theses workflow. This: * fix the issue of package not being able to be built because of rate limitation on other resources. * Reduce the chances of random errors. * Reduce the total CI times requirement. * Do not attempt to run the droplet test if the package building phase fail. (Previously all the package build were launched in parallel which mean they all failed unecessaryely) * Make it less costly and faster to run the failed jobs With theses chance the number of CI failure reduce greatly. And the cause of failure is more clear * CI: Wait till an ipv4 on droplet before proceeding It might be a change in the Digital Ocean API but it return before the network is setup and with empty setup info it didn't seem to occur befor so it might be an API change from their part * CI: Ensure we reuse the previously calculated IP * CI: Ensure we use the public IP v4 of the Droplet Digital Ocean droplet always have a private IP in addition to the public one. The API return them in random order so the CI job occasionally tried to use the internal one and failed. * CI: Merge the runtime test workflow with the package workflow Same operation as moving the Droplet workflow, we reuse the already build package. The resilience and speed advantage are the sames and add up. * CI: Remove the deleted workflow, update doc Rename the main workflow field Document more * improve doc * CI: Do not stop if Export aleph logs command fail * CI: Print commands run in install aleph step for debug * CI: Do not fetch whole repository where not needed In the run_on_droplet job we only require the .github/scripts dir * CI: Fix apt broken progress bar output
Previous way was preventing them from working inside tests
set up scaffolding to test reservation system
Jira: ALEPH-421
Part of Jira ALEPH-421
- Add supports_x_vga flag to QemuGPU model - Add system check to detect if a GPU supports x-vga feature - Only add x-vga parameter to QEMU command when supported - Fall back gracefully when professional GPUs don't support x-vga 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
- Test QemuVM device args generation with and without x-vga support - Test x-vga detection process in AlephQemuInstance - Test error handling for GPU x-vga detection - Test configuration flow with x-vga support detection 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
- Simplify x-vga support detection by using the GPU device class from lspci - Class 0300 (VGA compatible controller) supports x-vga - Class 0302 (3D controller) does not support x-vga - Remove the complex subprocess-based detection method in favor of this simpler approach - Update unit tests to test the device class-based detection This approach is faster, more reliable, and doesn't require running test QEMU commands. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
The test was trying to set attributes on HostGPU objects, but the HostGPU class doesn't have the supports_x_vga field. Changed to use GpuDevice which has the required field to fix the test. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
When aleph-vm-supervisor restarts, it properly loads existing VM interfaces but doesn't re-add their IPv6 ranges to the ndppd proxy configuration. This change explicitly re-adds ndp_proxy rules for existing interfaces during VM recovery. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
…t ndppd service for every VM.
of vm when calculate available disk space Jira: ALEPH-420 This will help the scheduler scheduler more properly the available resource on CRN Alternative approach to #780 as Sparse file were not properly created.
Don't call pyaleph if there is no execution for user Remove debug executions
The vm-connector service used to run with obsolete versions of the aleph-sdk and Python. Solution: Use the latest the version of the SDK and pin the version of dependencies.
olethanh
force-pushed
the
hoh-update-vm-connector
branch
from
03b780b to
f8eab86
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The vm-connector service used to run with obsolete versions of the aleph-sdk and Python.
Solution: Use the latest the version of the SDK and pin the version of dependencies.