BMP test rework #17306
Open
BMP test rework #17306
Conversation
This file contains 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
pguibert6WIND
changed the title
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
2 times, most recently
from
9c1b97b
to
9c91da3
Compare
ton31337
reviewed
Nov 4, 2024
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
2 times, most recently
from
77e03e2
to
fcdaf71
Compare
|
ci:rerun |
done |
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
2 times, most recently
from
4a1c38d
to
f68a64e
Compare
ton31337
reviewed
Nov 18, 2024
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
from
2ee8f99
to
c8381ce
Compare
ton31337
reviewed
Nov 20, 2024
| tgen.start_topology() | ||
|
|
||
| if DEBUG_PCAP: | ||
| tgen.gears["r1"].run("rm /tmp/bmp.pcap") |
There was a problem hiding this comment.
IMO, this should be something more granular, because if you run this in parallel, it won't work... Same for tcpdump below.
There was a problem hiding this comment.
DEBUG_PCAP is never enabled when running tests if this is what you fear.
Eventually, I can add a comment to say to not enable it while in CIP.
There was a problem hiding this comment.
The concern here is that you are writing to a arbitrary bit of the tmp directory when the FRR topotests already have a well established system in place where capturing log files exist in relation to what is going on. Let's try to keep this clean and consistent so that the next person doesn't add something to plain old /tmp and cause a real problem.
There was a problem hiding this comment.
could commit 4118279 help ?
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
from
4118279
to
ea42a86
Compare
|
Can we fix this? |
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
from
ea42a86
to
634ea31
Compare
done |
The bgp_bmp and bgp_bmp_vrf tests use similar routines to test the bmp, but are duplicates. Gather the bgp_bmp and the bgp_bmp_vrf tests in a single bgp_bmp folder. - Create a bgpbmp.py library under the bgp_bmp test folder - The bgp_bmp and bgp_bmp_vrf test are renamed to bgp_bmp_1 and bgp_bmp_2 test. - Maintain separate folder for config and output results. Adapt the bgp_bmp library accordingly. - The json output for bgp_bmp_2 test has no referenc to hostame. Signed-off-by: Philippe Guibert <[email protected]>
When configuring the bgp_bmp test with an additional
peer that sends empty AS-PATH, the bmp collector is stopping:
> [2024-10-28 17:41:51] Finished dissecting data from ('192.0.2.1', 33922)
> [2024-10-28 17:41:52] Data received from ('192.0.2.1', 33922): length 195
> [2024-10-28 17:41:52] Got message type: <class 'bmp.BMPRouteMonitoring'>
> [2024-10-28 17:41:52] unpack_from requires a buffer of at least 2 bytes for unpacking 2 bytes at offset 0 (actual buffer size is 0)
> [2024-10-28 17:41:52] TCP session closed with ('192.0.2.1', 33922)
> [2024-10-28 17:41:52] Server shutting down on 192.0.2.10:1789
The parser attempts to read an empty AS-path and considers the length
value as a length in bytes, whereas RFC mentions this value as
definining the number of AS-PAths. Fix this in the parser.
Signed-off-by: Philippe Guibert <[email protected]>
Use unified configuration procedure. Signed-off-by: Philippe Guibert <[email protected]>
to help identify the file type. And apply black. Signed-off-by: Louis Scalbert <[email protected]>
Multiple BMP tests can run in parallel but, when one instance ends, it kills the BMP server process of all BMP tests. Save the PID of a BMP server and only kill it at the end. Link: FRRouting#17465 Fixes: 875511c ("topotests: add basic bmp collector") Signed-off-by: Louis Scalbert <[email protected]>
DEBUG_PCAP can be set True to manually enable pcap debugging when running bmp tests. Save bmp pcap in logdir (ie. /tmp/topotests/bgp_bmp.bgp_bmp_X/ instead of /tmp. Signed-off-by: Louis Scalbert <[email protected]> Signed-off-by: Philippe Guibert <[email protected]>
pguibert6WIND
force-pushed
the
bmp_test_factorise_plus_fix
branch
from
634ea31
to
c01ffee
Compare
|
ci:rerun bfd test fail dont know why |
ton31337
approved these changes
Nov 27, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
See individual commits:
The BMP tests use a common library that is duplicate in bgp_bmp and bgp_bmp_vrf test.
Gather the two tests in one, and benefit from the same code.
When testing with an iBGP peering, a fix has been found in current BMP collector implementation regarding the read of the BGP as path attribute. Fix it too.
Protection against killing of undesirable bmp collector when testing multiple bmp tests in parallel: BMPServer in topotests should not
pkill -f bmpserver#17465