[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13778257#comment-13778257 ] Junping Du commented on HDFS-5208: -- Thanks Colin for review the patch again. The Jenkins test is finished and good now. Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch, HDFS-5208-v2.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776086#comment-13776086
]
Junping Du commented on HDFS-5208:
--
Hi Colin, I updated my new comments in HDFS-5237 and thanks for your comments
there. IMO, HDFS-5237 shouldn't be a blocking jira for this as no registration
name (only IP) going to cache backed by CachedDNSToSwitchMapping because of
following code in DatanodeManager.resolveNetworkLocation (DatanodeID).
{code}
if (dnsToSwitchMapping instanceof CachedDNSToSwitchMapping) {
names.add(node.getIpAddr());
} else {
names.add(node.getHostName());
}
{code}
What do you think?
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776547#comment-13776547
]
Colin Patrick McCabe commented on HDFS-5208:
{code}
- dnsToSwitchMapping.reloadCachedMappings();
+ ListString invalidNodeNames = new ArrayListString(1);
+ // clear cache for nodes in IP or Hostname
+ invalidNodeNames.add(nodeReg.getIpAddr());
+ invalidNodeNames.add(nodeReg.getHostName());
+ dnsToSwitchMapping.reloadCachedMappings(invalidNodeNames);
{code}
Can we also add something like this?
{code}
+ invalidNodeNames.add(nodeReg.getPeerHostName());
{code}
It seems like the datanode could be known by any one of those three: IP
address, registration name, or hostname.
After that change, I don't see any reason why this shouldn't work. What kind
of testing have you done?
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13777004#comment-13777004 ] Junping Du commented on HDFS-5208: -- Thanks Colin for review. v2 patch incorporate your comments. For test, it works well to clean up cache for node (in registration name) registering with fault topology in TestNetworkTopology. Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch, HDFS-5208-v2.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13777113#comment-13777113
]
Hadoop QA commented on HDFS-5208:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12604930/HDFS-5208-v2.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 5 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:green}+1 findbugs{color}. The patch does not introduce any new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-common-project/hadoop-common hadoop-hdfs-project/hadoop-hdfs
hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-HDFS-Build/5032//testReport/
Console output: https://builds.apache.org/job/PreCommit-HDFS-Build/5032//console
This message is automatically generated.
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch, HDFS-5208-v2.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13773076#comment-13773076 ] Junping Du commented on HDFS-5208: -- Filed HDFS-5237 to track node registration name issue and upload with a demo patch. However, as demo patch shows, lots of tests are affected by ignoring specified host name in MiniDFSCluster. Hi [~cmccabe], from your above comments it seems that we had some discussions before on fake up having different hostnames in simpler ways. It is great if you can share more info on this. Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13772100#comment-13772100 ] Colin Patrick McCabe commented on HDFS-5208: Sounds good to me. Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13770910#comment-13770910
]
Junping Du commented on HDFS-5208:
--
Hi Colin, I think you are right that DatanodeID is created in DN heartbeat to
NN for registration and its hostName comes from conf of dfs.datanode.hostname
which can be any style but DNS name if this config is not setting.
However, following code in resolveNetworkLocation() called by
DatanodeManager.registerDatanode() make only IPs are cached through DN
registration. Isn't it?
{code}
if (dnsToSwitchMapping instanceof CachedDNSToSwitchMapping) {
names.add(node.getIpAddr());
} else {
names.add(node.getHostName());
}
{code}
Actually, now I am worrying about non-cached case, as even topology script can
resolve user-specified hostName to correct network location (rack) properly
and use it to register into networktopology tree. Later, it still need to
resolve topology based on nodes' IP (like in
DatanodeManager.sortLocatedBlocks()) which means script must contains both
user-specified hostName and IP for each node. IMO, This is really unnecessary
and confusing. Thoughts?
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13771504#comment-13771504
]
Colin Patrick McCabe commented on HDFS-5208:
Unfortunately, {{DatanodeID#getHostName}} doesn't actually return the hostname.
It returns {{DatanodeID#hostName}}, which is either the registration name (if
it was specified) or the hostname (if it was not.)
I think this feature is mainly used in unit tests to fake up having different
hostnames-- something we could probably do this in a much simpler way. We've
discussed creating a JIRA to remove it before-- maybe it's time.
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13771529#comment-13771529 ] Junping Du commented on HDFS-5208: -- Thanks for comments, Colin! bq. Unfortunately, DatanodeID#getHostName doesn't actually return the hostname. It returns DatanodeID#hostName, which is either the registration name (if it was specified) or the hostname (if it was not.) So, we are saying the same thing - DatanodeID is created in DN heartbeat to NN for registration and its hostName comes from conf of dfs.datanode.hostname which can be any style but DNS name if this config is not setting. - Isn't it? :) bq. I think this feature is mainly used in unit tests to fake up having different hostnames-- something we could probably do this in a much simpler way. We've discussed creating a JIRA to remove it before-- maybe it's time. I agree. Will file separated Jira and work on it later. So we may just do simply two things below: 1. remove config of dfs.datanode.hostname and all usage place. 2. make sure DatanodeID#hostname is its hostname (DNS name). Anything else to address? Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13769605#comment-13769605
]
Colin Patrick McCabe commented on HDFS-5208:
If you look in DatanodeID, a base class of DatanodeRegistration, you'll find
this:
{code}
private String ipAddr; // IP address
private String hostName; // hostname claimed by datanode
private String peerHostName; // hostname from the actual connection
{code}
{{ipAddr}} is the IP address as a string. {{hostName}} is the registration
name. {{peerHostName}} is the hostname.
The registration name can't be resolved by DNS. In fact, it is completely
fake. But will be added to {{DNSToSwitchMapping}} if a datanode asks for it to
be added.
So really you need to clear all of them, plus all of them including a colon and
port.
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13768510#comment-13768510 ] Colin Patrick McCabe commented on HDFS-5208: It's a little more complicated than just clearing IP address and hostname. Nodes can be stored by registration name, hostname:port, hostname, IP address, IP address:port. It would be nice to file a JIRA to get rid of registration name. Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[ https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13769013#comment-13769013 ] Junping Du commented on HDFS-5208: -- Thanks for comments, Colin! I checked again on all callers of DNSToSwitchMapping.resolve() but didn't find anywhere to resolve a registration name (the only way to fill the cache). Also, DN registration will call resolveNetworkLocation() in DatanodeManager which only resolve IpAddr or Hostname, Am I missing something here? Only clear network location cache on specific nodes if invalid NetworkTopology happens -- Key: HDFS-5208 URL: https://issues.apache.org/jira/browse/HDFS-5208 Project: Hadoop HDFS Issue Type: Improvement Reporter: Junping Du Assignee: Junping Du Attachments: HDFS-5208-v1.patch After HDFS-4521, once a DN is registered with invalid networktopology, all cached rack info in DNSToSwitchMapping will be cleared. We should only clear cache on specific nodes. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HDFS-5208) Only clear network location cache on specific nodes if invalid NetworkTopology happens
[
https://issues.apache.org/jira/browse/HDFS-5208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13767653#comment-13767653
]
Hadoop QA commented on HDFS-5208:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12603196/HDFS-5208-v1.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 5 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:green}+1 findbugs{color}. The patch does not introduce any new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-common-project/hadoop-common hadoop-hdfs-project/hadoop-hdfs
hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-client
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-HDFS-Build/4974//testReport/
Console output: https://builds.apache.org/job/PreCommit-HDFS-Build/4974//console
This message is automatically generated.
Only clear network location cache on specific nodes if invalid
NetworkTopology happens
--
Key: HDFS-5208
URL: https://issues.apache.org/jira/browse/HDFS-5208
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Junping Du
Assignee: Junping Du
Attachments: HDFS-5208-v1.patch
After HDFS-4521, once a DN is registered with invalid networktopology, all
cached rack info in DNSToSwitchMapping will be cleared. We should only clear
cache on specific nodes.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
