Just a note about the CM API: From what I can tell, the API doesn't bring more parameters back that what you can see in the Configuration tab for each app/service. Being able to see them in the INFO logs was exactly what was needed.  However, it would be nice to be able to use the API to get the same info as the INFO logs provides.  I could see some automation opportunities in the future... ... View more

Not yet but I have a good bit of experience writing Python code using the CM API.  I'll look into it.  If there's anything useful, it might be helpful to post my findings here... ... View more

Hey Ben,   One very key thing you mentioned was the fact that the INFO logs write out the parameter settings.  Not sure if that includes defaults as I have not researched it yet but this is extremely useful as I may not have enough time to figure out why the servlets/services are not running or are not available but being able to see the parameter settings this way is more then sufficient.   For anyone needing a way to quickly see the parameters separated from the rest of the log information, you can use this egrep pattern I put together.  Hasn't been thoroughly tested but works so far: egrep '^--\w+=\w?+' /var/log/impalad/impalad.INFO   This command extracted 229 parameters from our INFO log file.  Sounds like a pretty complete list??? ... View more

Hey Ben,   >most services expose a "/conf" servlet in their WebUI which gives you the most complete set of actually used parameters. This should be the most promising source of thruth. Impala has a similar Servlet with path "/varz". I've been looking for this /varz path for Impala and cannot find the correct path.  Can you give me the port and path that it should be on?  Unless it's a name that is given during setup and it's different than "/varz"???     > The instance process view of Cloudera Manager shows you the actually distributed config files - which often helps a lot but does not include default values. You can reach it from a service (e.g. Impala) by clicking on "Instances" -> (the instance you want to see, e.g. Impala Catalog Server on a node) -> Processes I've seen this but I don't see anything that comes close to what I need.  This is the list of Impala related config files: impala-conf/fair-scheduler.xml impala-conf/llama-site.xml impala-conf/sentry-site.xml impala-conf/log4j.properties impala.keytab impala-conf/.htpasswd impala-conf/impalad_flags The only thing that comes close is impala-conf/impalad_flags file and that has no link.   Thanks!   ... View more

>"Your cluster does sound unhappy"   LOL.  I'd say more like pi$$ed.  🙂   > If it's a JVM issue, we've seen in some cases that increasing heap sizes helps. Setting ipc.client.rpc-timeout.ms to 60000 I'd say it's more like a set of overloaded namenodes.  And, according to the research I've done so far, for another problem, I have a theory that I can use a more up to date GC and it should increase the performance and reduce the number of Zookeeper failures we have.  Our ZK failures are sporadic and happen every 2 to 3 days.  Sometimes more sometimes less.  Moving ZK to separate nodes is not an option at this point and I'm doing all I can to try to reduce the number of failures short of moving the service. I'll check our settings on this and see if we can do one or both.  I suspect we have increased JVM heap already, but not sure?   >We've also seen the file handle cache that got enabled by default in CDH5.15 help a lot in reducing namenode load, I assume this is available before this version but was not enabled by default???  I'll look it up and see...   > I agree 100%. I think whoever named it was either overly optimistic and assumed there wouldn't be a significant gap in time, or it was named from the point of view of the code rather than the external system So, my question is, is there an indicator in the Query Details that indicates something was returned?  I know I get an initial set of results back.  Without that "fetch" metric meaning what the word actually says, I don't know what indicates how long it took to get the first set of records back???   Back to the original issue... Given that the issue appears to be the last query issued in Hue tends to show up as still executing 2.2 hours later and has already returned a count almost immediately.  Obviously, the parameters for idle timeouts for sessions and queries is not marking the session as closed.  Therefore appearing to still be executing: Is this causing resource issues because the session is being held open and appearing to be still executing?  I would assume so as it is waiting on a fetch of subsequent sets of records??? What parameter(s) will close the session from the last query executed?   Just to let you know, I've come in late to the game and am still learning CM and Cloudera Manager.  I understand a lot but with 1000s of parameters for all the apps and an ample supply of problems, it'll take a while.  🙂   Thanks for all your help.  It is nice to have a response on this forum.  The last couple posts were not very useful.  We do have a service contract and although I am one of the 3 admins, they are working on adding me to the support contract so I can put in tickets and get support directly. Until then, I appreciate the help! ... View more

Cool.   It's hard to believe a count of 1000 records is taking 2.2 hours.  So, I closed the session and did not see the mentioned "Released admission control resources" value.     It would be very helpful to know why this did not show up and what needs to be done to release the resources... BTW, it's not holding up queries from getting admitted as far as I can tell.  We ran into a problem where it did not have enough memory to allocate to a query and returned an error.  That's what got me started down this road in the first place.   When you say "after the first row was fetched" what constitutes a fetch?  When I execute a query, I get returned results and I see the records.  I assume the fetch is automatic up to a certain number as it will page the rest of them.  To me, when I see records, a fetch has taken place??? This makes zero sense: "one quirk of the "first row fetched" event is that it tracks when the row was requested, not when it was returned)" as the very word fetch means it got a result and pulled it back for viewing.  Why would the word "fetch" be used in place of the word "requested?"   I get a result back of 1000.  There is no more to fetch from the query.  That's it.  It appears that the fetch occurred, I got my result and I'm completely done with the query.  That's when the resources should be released.   It sounds like you're saying the interface in Impala is showing incorrectly and has been fixed?  This would be the whole problem, correct?   Thanks! ... View more

I apologize for missing your request to get the query profile.  I can send you a new one as this is easily duplicatable.  I have already re-executed a query to see if the profile was that different... Now, I re-executed the same query which is a count and returns 1 row.  This query will show in the Impala Query monitor screen as Executing indefinitely.  I can close my session or re-execute another query and it will mark it as Finished.   Here's the same one 45 minutes into waiting on it too close.  It will not close unless I do something to make it close.  I waited over 2 hours the last time I tried and it still said it was Executing: Query: select count(*) from web_logs   Query Info Query ID: 184e512db38f7215:e361d04d00000000 User: <user> Database: default Coordinator: <host> Query Type: QUERY Query State: FINISHED Start Time: Aug 16, 2019 1:24:05 PM Duration: 47m, 43s Rows Produced: 1 Admission Result: Admitted immediately Admission Wait Time: 0ms Bytes Streamed: 64 B Client Fetch Wait Time: 6.59s Client Fetch Wait Time Percentage: 0 Connected User: hue/<host>@<domain> Delegated User: <user> Estimated per Node Peak Memory: 52.0 MiB File Formats: TEXT/NONE HDFS Average Scan Range: 101.1 KiB HDFS Bytes Read: 404.5 KiB HDFS Scanner Average Read Throughput: 1.4 GiB/s Impala Version: impalad version 2.11.0-cdh5.14.2 RELEASE (build ed85dce709da9557aeb28be89e8044947708876c) Network Address: <IP>:<Port> Node with Peak Memory Usage: <server>:<port> Out of Memory: false Per Node Peak Memory Usage: 196.4 KiB Planning Wait Time: 18ms Planning Wait Time Percentage: 0 Pool: root.<user> Query Status: OK Session ID: 7744f8b1bd92eb67:c892fbb3fbb81293 Session Type: HIVESERVER2 Statistics Corrupt: false Statistics Missing: true Threads: CPU Time: 75ms Threads: CPU Time Percentage: 50 Threads: Network Receive Wait Time: 32ms Threads: Network Receive Wait Time Percentage: 22 Threads: Network Send Wait Time: 1ms Threads: Network Send Wait Time Percentage: 1 Threads: Storage Wait Time: 40ms Threads: Storage Wait Time Percentage: 27 Threads: Total Time: 150ms Download Profile... Query Timeline Query submitted: 49.51us (49.51us) Planning finished: 18ms (18ms) Submit for admission: 19ms (733.33us) Completed admission: 19ms (134.25us) Ready to start on 4 backends: 19ms (521.71us) All 4 execution backends (5 fragment instances) started: 22ms (2ms) Rows available: 72ms (50ms) First row fetched: 6.66s (6.59s) Planner Timeline Analysis finished: 16ms (16ms) Value transfer graph computed: 16ms (27.31us) Single node plan created: 16ms (385.49us) Runtime filters computed: 16ms (54.11us) Distributed plan created: 16ms (24.85us) Lineage info computed: 16ms (26.54us) Planning finished: 17ms (467.37us)   Is this sufficient?     ... View more

Hi Tim,   This would be a good time to mention were are on this version of CM: CDH-5.14.2-1.cdh5.14.2.p0.3   Are you telling me that people that are executing a query in Hue, monitor the query in CM ---> Impala ---> Queries and after the idle timeouts are up, then the query shows it is not running anymore? Try this: Execute a query in Hue like a count on a table and monitor the status of the Impala query in CM.  Let me know if it actually times out. I discovered that the last statement executed in Hue will stay there forever or until you close the session or execute another statement. So, either of these parameters does not stop the session after the timeout period as they say they are supposed to do: -idle_session_timeout=3600 -idle_query_timeout=3600 Look at the attachment (apparently I can't past images in as I exceeded 100,000 characters??? This is actually how fast the results were returned and the query finished: First row fetched: 838ms (743ms)   I assume you have a test bed you can set the setting to a much lower number to see if it is still showing as running in the Impala query monitor tool in CM.  We are having sporadic resource issues because these resources are not being released.  The more people that run a query and don't purposely close out the session, the more resources are being consumed. So, all the parameters I mentioned before don't help a bit?  It's all up to these 2 parameters?  Maybe they are not being loaded???  I was looking for a way to hit the site and get all current parameter settings???  I know Hue does not have an API, but is there a way to see if it actually is loading those parameters?   Thanks for your help! ... View more

It's a bit late in the game but I'm running into the same problem where the query appears to be running for hours and the first row fetched in in seconds.  This means it is not actually running although the list of Impala queries says it is.  As previously stated by a poster, the user did not close the session.  I have just noticed that the last query executed will hold the query in a running state.  Once another query is executed or the session closed, it will release the resources and mark the query as finished. I have another post about this same issue.  Neither one of these 2 parameters have helped: -idle_session_timeout=1500 -idle_query_timeout=1500   So, my conclusion is the documentation is not accurate in what it says about these parameters or there's a bug as of 08/16/2019???   If you find out how to close a session on a query automatically using a parameter, let me know... ... View more