Display large number of FDB entries fast from state_db and appl_db #968

JiangboHe · 2020-06-28T09:46:07Z

- What I did
I change the mothod by showing mac table from state FDB_TABLE to display large number of fdb tables faster.
The time cost to display change from 50 seconds to 2 seconds for 8000 fdb tables or 20 seconds to 1 seconds for 5000 fdb tables.

- How I did it
The key search and judge of fdb tables in asic table is too complex. It will cost 20 seconds for 5000 fdb tables or 50 seconds for 8000 fdb tables. It is not good for users or script running.

- How to verify it
Learn thounds of fdb tables and show them with fdbshow

- Previous command output (if the output of a command-line utility has changed)

- New command output (if the output of a command-line utility has changed)

renukamanavalan · 2020-06-29T23:01:13Z

retest this please

prsunny · 2020-06-30T04:27:18Z

@anilkpandey , can you take a look? There was a performance optimization done as part of #529, and it is a different approach.

JiangboHe · 2020-06-30T10:32:13Z

I see the improvement from 10 minutes to 17-20 seconds in Layer 2 Forwarding Enhancements. What I do is to show 8K mac in 2 seconds.

prsunny · 2020-06-30T13:51:34Z

retest this please

JiangboHe · 2020-07-02T07:35:01Z

I have test again.
It took 2s instead of 30s for displaying 8K fdb tables(or 13s for 5k, 7s for 3k).

prsunny · 2020-07-02T18:40:52Z

I have test again.
It took 2s instead of 30s for displaying 8K fdb tables(or 13s for 5k, 7s for 3k).

ok, @qiluo-msft , can you take a look as well? improvement looks good to me

scripts/fdbshow

qiluo-msft · 2020-07-03T01:37:14Z

scripts/fdbshow

@@ -93,7 +93,24 @@ class FdbShow(object):
        self.bridge_mac_list.sort(key = lambda x: x[0])
        return

-
+    def fetch_state_fdb_data(self):


fetch_state_fdb_data [](start = 8, length = 20)

fetch_state_fdb_data [](start = 8, length = 20)

@kcudnik Please help check.
The StateDB FDB is learned from syncd notification. So is it the same as AsicDB, for example, how about static fdb entries programmed? #Closed

yes, asic db learns those from fdb notification, static entries are created by user explicitly, so those comes from OA

Yes, It does not process static fdb tables. It is a problem. #Closed

If you cannot find static FDB entries in StateDB, it will be inside AppDB. FYI if you want to further explore.

In reply to: 450189315 [](ancestors = 450189315)

So, thats a good reason to have an option to display from state_db based on need. By default, and for small number of entries, it can use the existing option. But for large number, user can specify to display from state_db

It is a good idea. I will add the option for the command.

I have add option "-f" to for displaying large number of FDB entries fastly from state_db and appl_db.
The static FDB entries are also displayed from appl_db in fast option.

root@sonic:/home/admin# fdbshow -h
usage: fdbshow [-h] [-p PORT] [-v VLAN] [-f]

Display ASIC FDB entries

optional arguments:
-h, --help show this help message and exit
-p PORT, --port PORT FDB learned on specific port: Ethernet0
-v VLAN, --vlan VLAN FDB learned on specific Vlan: 1001
-f, --fast Display large number of FDB entries fastly
root@sonic:/home/admin#
root@sonic:/home/admin# fdbshow -f
No. Vlan MacAddress Port Type

1 100 00:00:00:00:00:01 Ethernet1 dynamic 2 100 11:00:00:00:00:01 Ethernet1 static 3 100 11:00:00:00:00:02 Ethernet1 static 4 100 11:00:00:00:00:03 Ethernet1 static

Total number of entries 4

qiluo-msft · 2020-08-24T21:44:34Z

scripts/fdbshow


-    def __init__(self):
+    def __init__(self,fast):


, [](start = 21, length = 1)

, [](start = 21, length = 1)

Add one space after , #Closed

qiluo-msft · 2020-08-24T21:45:16Z

scripts/fdbshow

        super(FdbShow,self).__init__()
        self.db = SonicV2Connector(host="127.0.0.1")
        self.if_name_map, \
        self.if_oid_map = port_util.get_interface_oid_map(self.db)
        self.if_br_oid_map = port_util.get_bridge_port_map(self.db)
-        self.fetch_fdb_data()
+        self.fast=fast


= [](start = 17, length = 1)

= [](start = 17, length = 1)

Add one blank before and after = #Closed

qiluo-msft · 2020-08-24T21:45:39Z

scripts/fdbshow

        super(FdbShow,self).__init__()
        self.db = SonicV2Connector(host="127.0.0.1")
        self.if_name_map, \
        self.if_oid_map = port_util.get_interface_oid_map(self.db)
        self.if_br_oid_map = port_util.get_bridge_port_map(self.db)
-        self.fetch_fdb_data()
+        self.fast=fast
+        if (self.fast is True):


Suggested change

if (self.fast is True):

if self.fast:

``` #Closed

qiluo-msft · 2020-08-24T21:48:26Z

scripts/fdbshow

@@ -139,10 +175,11 @@ def main():
                                     formatter_class=argparse.RawTextHelpFormatter)
    parser.add_argument('-p', '--port', type=str, help='FDB learned on specific port: Ethernet0', default=None)
    parser.add_argument('-v', '--vlan', type=str, help='FDB learned on specific Vlan: 1001', default=None)
+    parser.add_argument('-f', '--fast', action='store_true', help='Display large number of FDB entries fastly')


fastly [](start = 103, length = 6)

fastly [](start = 103, length = 6)

fast #Closed

Thank you. I have changed codes as reviewed.

qiluo-msft

As comments

JiangboHe · 2020-09-08T10:19:34Z

Hi, Do you know what's wrong with the check result "default — Build finished. No test results found." ?

qiluo-msft

LGTM. Please keep old commits in future if you iterate in one PR. We have some internal tools to compare between commits. If you force push, the tool will confuse and lose the context of existing comments. Thanks!

qiluo-msft · 2020-09-08T18:07:35Z

Retest this please

prsunny · 2020-09-08T19:00:21Z

retest this please

JiangboHe · 2020-09-09T03:57:16Z

Hi qiluo,
I have fix LGTM failure. But what should I do to fix the check result "default — Build finished. No test results found." ? I have tried by pushing a empty commit to rebuild, but get the same result.

prsunny · 2020-09-09T15:55:32Z

retest this please

qiluo-msft · 2020-09-09T20:28:10Z

08:56:41  Copied 323 artifacts from "vs » buildimage-vs-all" build number 420

This artifacts are wrong with old code, we already fixed in sonic-net/sonic-buildimage@d4fc8e5

liat-grozovik · 2020-11-05T16:56:41Z

@JiangboHe will this PR will be further handled and taken to 201911?

JiangboHe · 2020-11-24T07:11:10Z

retest this please

JiangboHe · 2020-11-25T03:21:17Z

@JiangboHe will this PR will be further handled and taken to 201911?
Yes, we found and fixed it in 201911 branch. It can also be merged to master.

JiangboHe · 2020-11-26T09:24:51Z

I am confused by the default failure after "retest this please". I have tried several times. Do you known what caused it?
The output of build console is below:
15:59:29 Adding one-line test results to commit status...
15:59:29 Setting status of 1e93983 to FAILURE with url https://sonic-jenkins.westus2.cloudapp.azure.com/job/common/job/sonic-utilities-build-pr/3236/ and message: 'Build finished. No test results found.'
15:59:30 hudson.remoting.ProxyException: java.io.IOException: No space left on device
15:59:30 at sun.nio.ch.FileDispatcherImpl.write0(Native Method)
15:59:30 at sun.nio.ch.FileDispatcherImpl.write(FileDispatcherImpl.java:60)
15:59:30 at sun.nio.ch.IOUtil.writeFromNativeBuffer(IOUtil.java:93)
15:59:30 at sun.nio.ch.IOUtil.write(IOUtil.java:65)
15:59:30 at sun.nio.ch.FileChannelImpl.write(FileChannelImpl.java:211)
15:59:30 at java.nio.channels.Channels.writeFullyImpl(Channels.java:78)
15:59:30 at java.nio.channels.Channels.writeFully(Channels.java:101)
15:59:30 at java.nio.channels.Channels.access$000(Channels.java:61)
15:59:30 at java.nio.channels.Channels$1.write(Channels.java:174)
15:59:30 at java.io.OutputStream.write(OutputStream.java:75)
15:59:30 at hudson.remoting.ProxyOutputStream$Chunk$1.run(ProxyOutputStream.java:255)
15:59:30 at hudson.remoting.PipeWriter$1.run(PipeWriter.java:158)
15:59:30 at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
15:59:30 at java.util.concurrent.FutureTask.run(FutureTask.java:266)
15:59:30 at hudson.remoting.SingleLaneExecutorService$1.run(SingleLaneExecutorService.java:131)
15:59:30 at hudson.remoting.InterceptingExecutorService$1.call(InterceptingExecutorService.java:72)
15:59:30 Caused: java.io.IOException: Pipe is already closed
15:59:30 at hudson.remoting.PipeWindow.checkDeath(PipeWindow.java:122)
15:59:30 at hudson.remoting.PipeWindow$Real.get(PipeWindow.java:220)
15:59:30 at hudson.remoting.ProxyOutputStream.write(ProxyOutputStream.java:124)
15:59:30 at hudson.remoting.RemoteOutputStream.write(RemoteOutputStream.java:108)
15:59:30 at org.apache.commons.io.IOUtils.copyLarge(IOUtils.java:2315)
15:59:30 at org.apache.commons.io.IOUtils.copy(IOUtils.java:2270)
15:59:30 at org.apache.commons.io.IOUtils.copyLarge(IOUtils.java:2291)
15:59:30 at org.apache.commons.io.IOUtils.copy(IOUtils.java:2246)
15:59:30 at hudson.FilePath.copyFrom(FilePath.java:987)
15:59:30 at hudson.plugins.copyartifact.CopyArtifact.copyOne(CopyArtifact.java:617)
15:59:30 Caused: java.io.IOException: Failed to copy file:/var/jenkins_home/jobs/vs/jobs/buildimage-vs-201911/builds/215/archive/target/docker-nat.gz to /data/johnar/workspace/common/sonic-utilities-build-pr/buildimage/target/docker-nat.gz
15:59:30 at hudson.plugins.copyartifact.CopyArtifact.copyOne(CopyArtifact.java:634)
15:59:30 at hudson.plugins.copyartifact.CopyArtifact.copy(CopyArtifact.java:577)
15:59:30 at hudson.plugins.copyartifact.CopyArtifact.perform(CopyArtifact.java:537)
15:59:30 at hudson.plugins.copyartifact.CopyArtifact.perform(CopyArtifact.java:475)
15:59:30 at org.jenkinsci.plugins.workflow.steps.CoreStep$Execution.run(CoreStep.java:80)
15:59:30 at org.jenkinsci.plugins.workflow.steps.CoreStep$Execution.run(CoreStep.java:67)
15:59:30 at org.jenkinsci.plugins.workflow.steps.SynchronousNonBlockingStepExecution.lambda$start$0(SynchronousNonBlockingStepExecution.java:47)
15:59:30 at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
15:59:30 at java.util.concurrent.FutureTask.run(FutureTask.java:266)
15:59:30 at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
15:59:30 at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
15:59:30 at java.lang.Thread.run(Thread.java:748)
15:59:30 Finished: FAILURE

prsunny · 2020-11-30T23:05:44Z

@JiangboHe , is this fix already in master? Just wondering why this is raised to 201911?

JiangboHe · 2020-12-01T02:41:50Z

@JiangboHe , is this fix already in master? Just wondering why this is raised to 201911?
No, we found and fixed it in 201911 branch. I can fixed it in master later, I have commit a pull request Azure/sonic-utilities/#1273 in master.

prsunny · 2020-12-01T02:54:49Z

ok, in that case, lets get the master PR first and cherry-pick to 201911. We can use this PR, if the cherry-pick has conflicts. For now, you can close this.

qiluo-msft · 2020-12-01T07:11:35Z

Close and favor the PR to master branch.

JiangboHe changed the base branch from master to 201911 June 28, 2020 10:00

jleveque requested a review from prsunny June 29, 2020 22:09

jleveque added the Enhancement label Jun 29, 2020

qiluo-msft reviewed Jul 3, 2020

View reviewed changes

scripts/fdbshow Show resolved Hide resolved

qiluo-msft reviewed Jul 3, 2020

View reviewed changes

scripts/fdbshow Show resolved Hide resolved

qiluo-msft requested a review from kcudnik July 3, 2020 01:36

qiluo-msft reviewed Jul 3, 2020

View reviewed changes

prsunny mentioned this pull request Jul 14, 2020

[show] Fix for 'show mac' command during DB flush #988

Closed

qiluo-msft reviewed Aug 24, 2020

View reviewed changes

qiluo-msft requested changes Aug 24, 2020

View reviewed changes

JiangboHe force-pushed the 201911 branch from d95f778 to 2278ad9 Compare August 25, 2020 09:00

JiangboHe requested a review from qiluo-msft September 8, 2020 08:50

JiangboHe force-pushed the 201911 branch from 2278ad9 to e42dfdf Compare September 8, 2020 09:03

JiangboHe changed the title ~~show mac table from state FDB_TABLE to display large number of fdb tables faster~~ Display large number of FDB entries fast from state_db and appl_db Sep 8, 2020

JiangboHe force-pushed the 201911 branch from e42dfdf to 8bb53cd Compare September 8, 2020 10:13

qiluo-msft approved these changes Sep 8, 2020

View reviewed changes

JiangboHe force-pushed the 201911 branch from 631c8a4 to 8bb53cd Compare November 24, 2020 07:41

Display large number of FDB entries fast from state_db and appl_db

1e93983

JiangboHe force-pushed the 201911 branch from 8bb53cd to 1e93983 Compare November 24, 2020 07:57

JiangboHe mentioned this pull request Nov 25, 2020

Display large number of FDB entries fast from state_db and appl_db #1273

Open

qiluo-msft closed this Dec 1, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Display large number of FDB entries fast from state_db and appl_db #968

Display large number of FDB entries fast from state_db and appl_db #968

JiangboHe commented Jun 28, 2020

renukamanavalan commented Jun 29, 2020

prsunny commented Jun 30, 2020

JiangboHe commented Jun 30, 2020

prsunny commented Jun 30, 2020

JiangboHe commented Jul 2, 2020

prsunny commented Jul 2, 2020 •

edited

Loading

qiluo-msft Jul 3, 2020 •

edited

Loading

kcudnik Jul 3, 2020

JiangboHe Jul 6, 2020 •

edited by qiluo-msft

Loading

qiluo-msft Jul 6, 2020

prsunny Jul 6, 2020

JiangboHe Jul 9, 2020

JiangboHe Aug 21, 2020

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

JiangboHe Aug 26, 2020

qiluo-msft left a comment

JiangboHe commented Sep 8, 2020

qiluo-msft left a comment

qiluo-msft commented Sep 8, 2020

prsunny commented Sep 8, 2020

JiangboHe commented Sep 9, 2020

prsunny commented Sep 9, 2020

qiluo-msft commented Sep 9, 2020

liat-grozovik commented Nov 5, 2020

JiangboHe commented Nov 24, 2020

JiangboHe commented Nov 25, 2020

JiangboHe commented Nov 26, 2020

prsunny commented Nov 30, 2020

JiangboHe commented Dec 1, 2020 •

edited

Loading

prsunny commented Dec 1, 2020

qiluo-msft commented Dec 1, 2020

Display large number of FDB entries fast from state_db and appl_db #968

Display large number of FDB entries fast from state_db and appl_db #968

Conversation

JiangboHe commented Jun 28, 2020

renukamanavalan commented Jun 29, 2020

prsunny commented Jun 30, 2020

JiangboHe commented Jun 30, 2020

prsunny commented Jun 30, 2020

JiangboHe commented Jul 2, 2020

prsunny commented Jul 2, 2020 • edited Loading

qiluo-msft Jul 3, 2020 • edited Loading

Choose a reason for hiding this comment

kcudnik Jul 3, 2020

Choose a reason for hiding this comment

JiangboHe Jul 6, 2020 • edited by qiluo-msft Loading

Choose a reason for hiding this comment

qiluo-msft Jul 6, 2020

Choose a reason for hiding this comment

prsunny Jul 6, 2020

Choose a reason for hiding this comment

JiangboHe Jul 9, 2020

Choose a reason for hiding this comment

JiangboHe Aug 21, 2020

Choose a reason for hiding this comment

qiluo-msft Aug 24, 2020 • edited Loading

Choose a reason for hiding this comment

qiluo-msft Aug 24, 2020 • edited Loading

Choose a reason for hiding this comment

qiluo-msft Aug 24, 2020 • edited Loading

Choose a reason for hiding this comment

qiluo-msft Aug 24, 2020 • edited Loading

Choose a reason for hiding this comment

JiangboHe Aug 26, 2020

Choose a reason for hiding this comment

qiluo-msft left a comment

Choose a reason for hiding this comment

JiangboHe commented Sep 8, 2020

qiluo-msft left a comment

Choose a reason for hiding this comment

qiluo-msft commented Sep 8, 2020

prsunny commented Sep 8, 2020

JiangboHe commented Sep 9, 2020

prsunny commented Sep 9, 2020

qiluo-msft commented Sep 9, 2020

liat-grozovik commented Nov 5, 2020

JiangboHe commented Nov 24, 2020

JiangboHe commented Nov 25, 2020

JiangboHe commented Nov 26, 2020

prsunny commented Nov 30, 2020

JiangboHe commented Dec 1, 2020 • edited Loading

prsunny commented Dec 1, 2020

qiluo-msft commented Dec 1, 2020

prsunny commented Jul 2, 2020 •

edited

Loading

qiluo-msft Jul 3, 2020 •

edited

Loading

JiangboHe Jul 6, 2020 •

edited by qiluo-msft

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

qiluo-msft Aug 24, 2020 •

edited

Loading

JiangboHe commented Dec 1, 2020 •

edited

Loading