[jira] [Created] (HBASE-11326) Use an InputFormat for ExportSnapshot
Matteo Bertozzi created HBASE-11326: --- Summary: Use an InputFormat for ExportSnapshot Key: HBASE-11326 URL: https://issues.apache.org/jira/browse/HBASE-11326 Project: HBase Issue Type: Improvement Components: snapshots Affects Versions: 0.99.0 Reporter: Matteo Bertozzi Assignee: Matteo Bertozzi Priority: Minor Fix For: 0.99.0 Use an InputFormat instead of uploading a set of input files to have a progress based on the file size -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: Anyone to have a quick look at 11313?
I took a look. While the patch is trivial enough, I think we should do something do something more drastic. We should deprecate/remove Reusable and Threadlocal PoolMaps for the rpc clients. Reusable PoolMap in conjunction with the code becomes resource unbounded and ThreadLocal is tricky to use (difficult to close or interrupt since we have to hunt down all the threads). For now we should deprecate the other two and once 1.0 gets branched we should remove Reusable and Threadlocal and possibly just get rid of the extra config infrastructure. Jon. On Tue, Jun 10, 2014 at 5:27 PM, Jean-Marc Spaggiari jean-m...@spaggiari.org wrote: Hi Konstantin, I'm talking about this one: https://issues.apache.org/jira/browse/HBASE-11313 Thanks, JM 2014-06-10 20:14 GMT-04:00 Konstantin Boudnik c...@apache.org: Seems to be a private one? Unless I am not in the correct group or something... On Tue, Jun 10, 2014 at 07:58PM, Jean-Marc Spaggiari wrote: Pretty trivial patch. Thanks, JM -- // Jonathan Hsieh (shay) // HBase Tech Lead, Software Engineer, Cloudera // j...@cloudera.com // @jmhsieh
[jira] [Created] (HBASE-11327) ExportSnapshot hit stackoverflow error when target snapshotDir doesn't contain uri
Demai Ni created HBASE-11327: Summary: ExportSnapshot hit stackoverflow error when target snapshotDir doesn't contain uri Key: HBASE-11327 URL: https://issues.apache.org/jira/browse/HBASE-11327 Project: HBase Issue Type: Bug Components: snapshots Affects Versions: 0.98.2 Reporter: Demai Ni Assignee: Demai Ni Priority: Minor Fix For: 0.99.0, 0.98.4 {code} $hbase org.apache.hadoop.hbase.snapshot.ExportSnapshot -snapshot snapshotT1_dn -copy-to /user/demai/backup1 Exception in thread main java.lang.StackOverflowError at java.util.regex.Pattern$Slice.match(Pattern.java:3490) at java.util.regex.Pattern$Start.match(Pattern.java:3066) at java.util.regex.Matcher.search(Matcher.java:1116) at java.util.regex.Matcher.find(Matcher.java:546) at org.apache.hadoop.conf.Configuration.substituteVars(Configuration.java:681) at org.apache.hadoop.conf.Configuration.get(Configuration.java:893) at org.apache.hadoop.fs.FileSystem.getDefaultUri(FileSystem.java:175) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:167) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:360) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:167) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:360) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:167) at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:360) {code} the following command will work with uri {code} hbase org.apache.hadoop.hbase.snapshot.ExportSnapshot -snapshot snapshotT1_dn -copy-to hdfs://hdtest014.svl.ibm.com:9000/user/demai/backup2 {code} The bug is the same as [Hadoop-9069|https://issues.apache.org/jira/browse/HADOOP-9069]. Since the hadoop jira has been sitting there for more than a year, use this jira for a local hbase fix for now. Many thanks for [~mbertozzi] help on this one. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Resolved] (HBASE-11313) RpcClient should allow Reusable pool option.
[ https://issues.apache.org/jira/browse/HBASE-11313?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jean-Marc Spaggiari resolved HBASE-11313. - Resolution: Won't Fix As Jon said. We will need a bigger modification. Will be addressed in a subsequent patch. RpcClient should allow Reusable pool option. Key: HBASE-11313 URL: https://issues.apache.org/jira/browse/HBASE-11313 Project: HBase Issue Type: Bug Affects Versions: 0.99.0, 0.98.3 Reporter: Jean-Marc Spaggiari Assignee: Jean-Marc Spaggiari Priority: Minor Attachments: HBASE-11313-v0-trunk.patch RpcClient.getPoolType check the pool choice against ThreadLocal and RoundRobin. However, we have a 3rd pooltype already in, Reusable. We should allow in this getPoolType check. -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: Anyone to have a quick look at 11313?
bq. we should deprecate the other two +1 On Wed, Jun 11, 2014 at 8:13 AM, Jonathan Hsieh j...@cloudera.com wrote: I took a look. While the patch is trivial enough, I think we should do something do something more drastic. We should deprecate/remove Reusable and Threadlocal PoolMaps for the rpc clients. Reusable PoolMap in conjunction with the code becomes resource unbounded and ThreadLocal is tricky to use (difficult to close or interrupt since we have to hunt down all the threads). For now we should deprecate the other two and once 1.0 gets branched we should remove Reusable and Threadlocal and possibly just get rid of the extra config infrastructure. Jon. On Tue, Jun 10, 2014 at 5:27 PM, Jean-Marc Spaggiari jean-m...@spaggiari.org wrote: Hi Konstantin, I'm talking about this one: https://issues.apache.org/jira/browse/HBASE-11313 Thanks, JM 2014-06-10 20:14 GMT-04:00 Konstantin Boudnik c...@apache.org: Seems to be a private one? Unless I am not in the correct group or something... On Tue, Jun 10, 2014 at 07:58PM, Jean-Marc Spaggiari wrote: Pretty trivial patch. Thanks, JM -- // Jonathan Hsieh (shay) // HBase Tech Lead, Software Engineer, Cloudera // j...@cloudera.com // @jmhsieh
Re: Anyone to have a quick look at 11313?
Thanks for looking at it Jon. I will close and open a new one for Reusable and ThreadLocal removal/deprecation. JM 2014-06-11 11:13 GMT-04:00 Jonathan Hsieh j...@cloudera.com: I took a look. While the patch is trivial enough, I think we should do something do something more drastic. We should deprecate/remove Reusable and Threadlocal PoolMaps for the rpc clients. Reusable PoolMap in conjunction with the code becomes resource unbounded and ThreadLocal is tricky to use (difficult to close or interrupt since we have to hunt down all the threads). For now we should deprecate the other two and once 1.0 gets branched we should remove Reusable and Threadlocal and possibly just get rid of the extra config infrastructure. Jon. On Tue, Jun 10, 2014 at 5:27 PM, Jean-Marc Spaggiari jean-m...@spaggiari.org wrote: Hi Konstantin, I'm talking about this one: https://issues.apache.org/jira/browse/HBASE-11313 Thanks, JM 2014-06-10 20:14 GMT-04:00 Konstantin Boudnik c...@apache.org: Seems to be a private one? Unless I am not in the correct group or something... On Tue, Jun 10, 2014 at 07:58PM, Jean-Marc Spaggiari wrote: Pretty trivial patch. Thanks, JM -- // Jonathan Hsieh (shay) // HBase Tech Lead, Software Engineer, Cloudera // j...@cloudera.com // @jmhsieh
Fork protobuf?
FYI. There's a fairly serious thread about forking protobuf happening over on HBASE-8. Should be socialized a little wider, I think.
[jira] [Created] (HBASE-11328) testMoveRegion could fail
Jimmy Xiang created HBASE-11328: --- Summary: testMoveRegion could fail Key: HBASE-11328 URL: https://issues.apache.org/jira/browse/HBASE-11328 Project: HBase Issue Type: Test Reporter: Jimmy Xiang Assignee: Jimmy Xiang Priority: Minor TestAssignmentManagerOnCluster#testMoveRegion could try to move a region to a server not online, and fail. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11329) Minor fixup of new blockcache tab number formatting
stack created HBASE-11329: - Summary: Minor fixup of new blockcache tab number formatting Key: HBASE-11329 URL: https://issues.apache.org/jira/browse/HBASE-11329 Project: HBase Issue Type: Bug Reporter: stack Priority: Trivial Counts are showing as MB/KB. Let me fix. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Resolved] (HBASE-11329) Minor fixup of new blockcache tab number formatting
[ https://issues.apache.org/jira/browse/HBASE-11329?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack resolved HBASE-11329. --- Resolution: Fixed Fix Version/s: 0.99.0 Assignee: stack Committed to master. Minor fixup of new blockcache tab number formatting --- Key: HBASE-11329 URL: https://issues.apache.org/jira/browse/HBASE-11329 Project: HBase Issue Type: Bug Reporter: stack Assignee: stack Priority: Trivial Fix For: 0.99.0 Attachments: 11329.txt Counts are showing as MB/KB. Let me fix. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11330) Deprecate ThreadLocal and Reusable from PoolMap
Jean-Marc Spaggiari created HBASE-11330: --- Summary: Deprecate ThreadLocal and Reusable from PoolMap Key: HBASE-11330 URL: https://issues.apache.org/jira/browse/HBASE-11330 Project: HBase Issue Type: Bug Reporter: Jean-Marc Spaggiari Assignee: Jean-Marc Spaggiari Reusable is not used anywhere and ThreadLocal is not recommend. See HBASE-11313 for other details. So let's remove Reusable and deprecate ThreadLocal. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11331) [blockcache] lazy block decompression
Nick Dimiduk created HBASE-11331: Summary: [blockcache] lazy block decompression Key: HBASE-11331 URL: https://issues.apache.org/jira/browse/HBASE-11331 Project: HBase Issue Type: Improvement Components: regionserver Reporter: Nick Dimiduk Assignee: Nick Dimiduk Maintaining data in its compressed form in the block cache will greatly increase our effective blockcache size and should show a meaning improvement in cache hit rates in well designed applications. The idea here is to lazily decompress/decrypt blocks when they're consumed, rather than as soon as they're pulled off of disk. This is related to but less invasive than HBASE-8894. -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: Timestamp resolution
Weirdly enough I find that I have to agree with Andrew. First, how do you get time in units smaller than a ms? Second clock skew becomes an issue. Third, which clock are you using? The client machine? The RS? And then how do you synchronize each of the RS to be within a ms of each other? Correct me if I’m wrong but NTP doesn’t give that close of a sync. Sorry, but really, not a good idea. If you want this… you can store the temporal data as a column. Time really is relative. On May 25, 2014, at 12:53 AM, Stack st...@duboce.net wrote: On Fri, May 23, 2014 at 5:27 PM, lars hofhansl la...@apache.org wrote: We have discussed this in the past. It just came up again during an internal discussion. Currently we simply store a Java timestamp (millisec since epoch), i.e. we have ms resolution. We do have 8 bytes for the TS, though. Not enough to store nanosecs (that would only cover 2^63/10^9/3600/24/365.24 = 292.279 years), but enough for microseconds (292279 years). Should we just store he TS is microseconds? We could do that right now (and just keep the ms resolution for now - i.e. the us part would always be 0 for now). Existing data must be in ms of course, so we'd grandfather that in, but new tables could store by default in us. We'd need to make this configurable both the column family level and client level, so clients could still opt to see data in ms. Comments? Too much to bite off? -- Lars I'm a fan. As Enis cites, HBASE-8927 has good discussion. No configuration I'd say. Just move to the new regime (though I suppose we should let you turn it off). I think it was Liu Shaohui (IIRC) who made a suggestion that had us put together ms and nanos under a synchronized block stamping the ts on Cells (left-shift the currentTimeMillis and fill in the bottom bytes with as much of the nanos as fits; i.e. your micros). Rather than nanos/micros, we could use a counter instead if a Cell arrives in the same ms. Would be costly having all ops go via one code block to get 'time' across cores and handlers. St.Ack
[jira] [Created] (HBASE-11332) Fix for metas location cache from HBASE-10785
Enis Soztutar created HBASE-11332: - Summary: Fix for metas location cache from HBASE-10785 Key: HBASE-11332 URL: https://issues.apache.org/jira/browse/HBASE-11332 Project: HBase Issue Type: Sub-task Reporter: Enis Soztutar Assignee: Enis Soztutar Fix For: hbase-10070 In HBASE-10785, we removed the invalidation of the cached location from patch v2 to v3. This results in a case where if there is a cached location for meta, it is not invalidated. Since we do a second check from cache for the location after acquiring the lock, this results in metas location to be wrongly cached forever resulting in clients blocking indefinitely. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11333) Remove deprecated class MetaMigrationConvertingToPB
Mikhail Antonov created HBASE-11333: --- Summary: Remove deprecated class MetaMigrationConvertingToPB Key: HBASE-11333 URL: https://issues.apache.org/jira/browse/HBASE-11333 Project: HBase Issue Type: Bug Components: master Affects Versions: 0.99.0 Reporter: Mikhail Antonov Assignee: Mikhail Antonov Priority: Trivial Fix For: 0.99.0 MetaMigrationConvertingToPB is marked deprecated and to be deleted next major release after 0.96. Is that the time? -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11334) Migrate to SLF4J as logging interface
jay vyas created HBASE-11334: Summary: Migrate to SLF4J as logging interface Key: HBASE-11334 URL: https://issues.apache.org/jira/browse/HBASE-11334 Project: HBase Issue Type: Improvement Reporter: jay vyas Migrating to new log implementations is underway as in HBASE-10092. Next step would be to abstract them so that the hadoop community can standardize on a logging layer that is easy for end users to tune. Simplest way to do this is use SLF4j APIs as the main interface and binding/ implementation details in the docs as necessary. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11335) Fix the TABLE_DIR param in TableSnapshotInputFormat
deepankar created HBASE-11335: - Summary: Fix the TABLE_DIR param in TableSnapshotInputFormat Key: HBASE-11335 URL: https://issues.apache.org/jira/browse/HBASE-11335 Project: HBase Issue Type: Bug Components: mapreduce, snapshots Affects Versions: 0.98.3, 0.96.2 Reporter: deepankar In class *TableSnapshotInputFormat* or *TableSnapshotInputFormatImpl* in the function {code} public static void setInput(Job job, String snapshotName, Path restoreDir) throws IOException { {code} we are setting restoreDir (temporary root) to tableDir {code} conf.set(TABLE_DIR_KEY, restoreDir.toString()); {code} The above parameter is used to get the InputSplits, especially for calculating favorable hosts in the function {code} Path tableDir = new Path(conf.get(TABLE_DIR_KEY)); ListString hosts = getBestLocations(conf, HRegion.computeHDFSBlocksDistribution(conf, htd, hri, tableDir)); {code} This will lead to returning a empty *HDFSBlocksDistribution*, as there is will be no directory with name as the region name from hri in the restored root directory, which will lead to scheduling of non local tasks. The change is simple in the sense, is to call the {code}FSUtils.getTableDir(rootDir, tableDesc.getTableName()) {code} in the getSplits function more discussion in the comments below https://issues.apache.org/jira/browse/HBASE-8369?focusedCommentId=14012085page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14012085 -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11336) Show region sizes in table page of master info server
Yi Deng created HBASE-11336: --- Summary: Show region sizes in table page of master info server Key: HBASE-11336 URL: https://issues.apache.org/jira/browse/HBASE-11336 Project: HBase Issue Type: Bug Components: Admin Affects Versions: 0.89-fb Reporter: Yi Deng Priority: Minor Show region sizes in table page of master info server -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Created] (HBASE-11337) Document how to create a table using Java
Misty Stanley-Jones created HBASE-11337: --- Summary: Document how to create a table using Java Key: HBASE-11337 URL: https://issues.apache.org/jira/browse/HBASE-11337 Project: HBase Issue Type: Bug Components: Admin, documentation Reporter: Misty Stanley-Jones Assignee: Misty Stanley-Jones Example code from [~jmspaggi] {quote} package com.cloudera.sa.hp.hbase.admin; import java.io.IOException; import org.apache.hadoop.hbase.HBaseConfiguration; import org.apache.hadoop.hbase.HColumnDescriptor; import org.apache.hadoop.hbase.HTableDescriptor; import org.apache.hadoop.hbase.TableName; import org.apache.hadoop.hbase.client.HBaseAdmin; import org.apache.hadoop.hbase.io.compress.Compression.Algorithm; import org.apache.hadoop.conf.Configuration; import static com.cloudera.sa.hp.hbase.Constants.*; public class CreateSchema { public static void createIfNotExist(HBaseAdmin admin, HTableDescriptor table) throws IOException { if (admin.tableExists(table.getName())) { admin.disableTable(table.getName()); admin.deleteTable(table.getName()); } admin.createTable(table); } public static void main(String[] args) { /**/ /* Create application schema. */ /**/ Configuration config = HBaseConfiguration.create(); config.set(hbase.zookeeper.quorum, 192.168.56.102); // Here we are running zookeeper locally try { final HBaseAdmin admin = new HBaseAdmin(config); HTableDescriptor table_assetmeta = new HTableDescriptor(TableName.valueOf(TABLE_ASSETMETA)); table_assetmeta.addFamily(new HColumnDescriptor(CF_DEFAULT).setCompressionType(Algorithm.GZ)); System.out.print(Creating table_assetmeta. ); admin.createTable(table_assetmeta); System.out.println( Done.); admin.close(); } catch (Exception e) { e.printStackTrace(); } } } {quote} -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: Fork protobuf?
Agreed, it's radical, but something arrived at by force On Wed, Jun 11, 2014 at 8:53 AM, Nick Dimiduk ndimi...@gmail.com wrote: FYI. There's a fairly serious thread about forking protobuf happening over on HBASE-8. Should be socialized a little wider, I think. -- Best regards, - Andy Problems worthy of attack prove their worth by hitting back. - Piet Hein (via Tom White)
Re: Fork protobuf?
I think we should do this - opening door for further optimizations. Cheers On Wed, Jun 11, 2014 at 8:57 PM, Andrew Purtell apurt...@apache.org wrote: Agreed, it's radical, but something arrived at by force On Wed, Jun 11, 2014 at 8:53 AM, Nick Dimiduk ndimi...@gmail.com wrote: FYI. There's a fairly serious thread about forking protobuf happening over on HBASE-8. Should be socialized a little wider, I think. -- Best regards, - Andy Problems worthy of attack prove their worth by hitting back. - Piet Hein (via Tom White)
Re: Thinking of branching for 1.0 by June 23
I would be happy to provide an overview of variation of this model that I've been using on different projects and companies. Please let me know if any is needed. Thanks, Cos On Tue, Jun 10, 2014 at 06:24PM, Enis Söztutar wrote: I think we can accept the patches for the zk abstraction. Let me link the parent issues together. For the git workflow, it would be good to hear from somebody with experience on different models (esp with the accumulo model). Enis On Tue, Jun 10, 2014 at 4:41 PM, Ted Yu yuzhih...@gmail.com wrote: I like Cos' idea. Cheers On Tue, Jun 10, 2014 at 4:34 PM, Konstantin Boudnik c...@apache.org wrote: I am +1 on b) because it will naturally allow for a continuation of 1.x development. In all honesty, I found that notorious git branching model works perfectly for such situations. One thing to mention: unlike http://nvie.com/posts/a-successful-git-branching-model/ it'll force a significant number of cherry-picking from the master (and SHA1 changes on such commits). Perhaps it might be a good time to reconsider what has been working ok for Hadoop on SVN and look into something that's more natural for Git branching? Cos On Tue, Jun 10, 2014 at 07:16PM, Jean-Marc Spaggiari wrote: For people voting, can you please put small comments regarding why you prefer a solution versus the other one? Just for knowledge sharing... Thanks, JM 2014-06-10 19:05 GMT-04:00 Mikhail Antonov olorinb...@gmail.com: I think jiras on ZK abstraction can still get committed (I'll make sure to have all non-trivial patches posted on RB for discussion to make sure we don't accidentally introduce any instability). On jiras. Under HBASE-10909: - HBASE-11069 (region merge transaction) is close to completion, just needs rebasing/merging, so we should have the new patch soon - HBASE-11072 (WAL splitting) - there's progress going on here, I think we're going to have patch up for reviews pretty soon. - HBASE-11073 (abstract Zk Watcher and listeners) - should have first patch up for review in a week or two Besides that, we should have HBASE-4495 (get rid of CatalogTracker) too. Further steps on abstraction (involving changing/simplifying the way we keep state in ZK) require coordination engine (as described in consensus design doc), which has been proposed in hadoop-common (for the time being I guess we can add this engine directly to HBase to speedup development?). Mikhail 2014-06-10 15:46 GMT-07:00 Stack st...@duboce.net: +1 on option b) On Tue, Jun 10, 2014 at 3:28 PM, Konstantin Boudnik c...@apache.org wrote: +1 on the #2. One question though: do you envision that the work around coordinated replication won't be able to go into branch-1 anymore? Its not done and it is far along with Mikhail making good progress. I'd be up for keeping up reviews and commit (if thats OK w/ you Mr. RM). How much you think could make 1.0 Cos/Mikhail? Which issues. St.Ack -- Thanks, Michael Antonov
[jira] [Created] (HBASE-11338) Expand documentation on bloom filters
Misty Stanley-Jones created HBASE-11338: --- Summary: Expand documentation on bloom filters Key: HBASE-11338 URL: https://issues.apache.org/jira/browse/HBASE-11338 Project: HBase Issue Type: Bug Components: documentation, Filters Reporter: Misty Stanley-Jones Assignee: Misty Stanley-Jones Ref Guide could use more info on bloom filters. -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: Timestamp resolution
The issues you cite are all orthogonal. We have client/RS time now, we have clock skew now, that is completely independent from the time resolution. I explained the need I saw for this before. Lemme include: On Fri, May 23, 2014 at 06:16PM, lars hofhansl wrote: The specific discussion here was a transaction engine doing snapshot isolation using the HBase timestamps, but still be close to wall clock time as much as possible. In that scenario, with ms resolution you can only do 1000 transactions/sec, and so you need to turn the timestamp into something that is not wall clock time as HBase understands it (and hence TTL, etc, will no longer work, as well as any other tools you've written that use the HBase timestamp). 1m transactions/sec are good enough (for now, I envision in a few years we'll be sitting here wondering how we could ever think that 1m transaction/sec are sufficient) :) The point is: Even if you had timestamp oracle (that can resolve ms and fill inside ms resolution with a counter), there'd be no way to use this as the HBase timestamp while being close to wall clock (so that TTL, etc, still works). So specifically I was not advocating an automatic higher time resolution (as far as I know that cannot be done reliably in Java across multiple cores). I was advocating allowing clients with access to a (perhaps, but not necessarily single threaded) timestamp oracle to store those timestamps and still make use of all HBase optimization (filtering HFiles, TTL, etc). -- Lars From: Michael Segel michael_se...@hotmail.com To: dev@hbase.apache.org Cc: lars hofhansl la...@apache.org Sent: Wednesday, June 11, 2014 2:03 PM Subject: Re: Timestamp resolution Weirdly enough I find that I have to agree with Andrew. First, how do you get time in units smaller than a ms? Second clock skew becomes an issue. Third, which clock are you using? The client machine? The RS? And then how do you synchronize each of the RS to be within a ms of each other? Correct me if I’m wrong but NTP doesn’t give that close of a sync. Sorry, but really, not a good idea. If you want this… you can store the temporal data as a column. Time really is relative. On May 25, 2014, at 12:53 AM, Stack st...@duboce.net wrote: On Fri, May 23, 2014 at 5:27 PM, lars hofhansl la...@apache.org wrote: We have discussed this in the past. It just came up again during an internal discussion. Currently we simply store a Java timestamp (millisec since epoch), i.e. we have ms resolution. We do have 8 bytes for the TS, though. Not enough to store nanosecs (that would only cover 2^63/10^9/3600/24/365.24 = 292.279 years), but enough for microseconds (292279 years). Should we just store he TS is microseconds? We could do that right now (and just keep the ms resolution for now - i.e. the us part would always be 0 for now). Existing data must be in ms of course, so we'd grandfather that in, but new tables could store by default in us. We'd need to make this configurable both the column family level and client level, so clients could still opt to see data in ms. Comments? Too much to bite off? -- Lars I'm a fan. As Enis cites, HBASE-8927 has good discussion. No configuration I'd say. Just move to the new regime (though I suppose we should let you turn it off). I think it was Liu Shaohui (IIRC) who made a suggestion that had us put together ms and nanos under a synchronized block stamping the ts on Cells (left-shift the currentTimeMillis and fill in the bottom bytes with as much of the nanos as fits; i.e. your micros). Rather than nanos/micros, we could use a counter instead if a Cell arrives in the same ms. Would be costly having all ops go via one code block to get 'time' across cores and handlers. St.Ack