bq. write old name only and within code we will have to map it with new name.
This is a viable approach. Even if we document upgrading to 1.2.3+, some users may not perform this step before upgrading to 2.0 On Tue, Dec 6, 2016 at 10:04 PM, Anoop John <[email protected]> wrote: > Ya that bug raised by Enis was very valid.. So this means when > rolling upgrade happens, if some of the other RSs with version <1.2.3 > , which is not having this fix, this issue might come up! > How to address then? Do we need to enforce a 1.2.3+ upgrade 1st and > then only 2.0 rolling upgrade? > > Or else we will need to fix in 2.0.. We write the new Comparator > class name in FFT (trunk code) To fix, we might have to write old > name only and within code we will have to map it with new name. It > will be ugly! It can be fixed only in 3.0 may be.. But that can make > the rolling upgrade story easier for users.. Just saying the > possibilities.. > > Thanks Ted to bring it up again. > > -Anoop- > > On Wed, Dec 7, 2016 at 10:04 AM, ramkrishna vasudevan > <[email protected]> wrote: > > I think when Enis reported the issue ( > > https://issues.apache.org/jira/browse/HBASE-16189), in rolling upgrade > > regions could move around. So a region created in 2.0 moved to a server > > with 1.x. > > > > Regards > > Ram > > > > > > On Wed, Dec 7, 2016 at 1:27 AM, Stack <[email protected]> wrote: > > > >> On Tue, Dec 6, 2016 at 10:19 AM, Ted Yu <[email protected]> wrote: > >> > >> > Looking at http://hbase.apache.org/book.html#executing.the.0.96. > upgrade > >> , > >> > there is step of running "bin/hbase upgrade -check" > >> > > >> > How about adding a sample hfile which contains > >> > CellComparator$MetaCellComparator > >> > so that the upgrade check can read and verify ? > >> > > >> > > >> > >> We don't support downgrade. Never have. > >> St.Ack > >> > >> > >> > >> > On Tue, Dec 6, 2016 at 8:50 AM, Ted Yu <[email protected]> wrote: > >> > > >> > > The build I used yesterday didn't include HBASE-16189 > >> > > <https://issues.apache.org/jira/browse/HBASE-16189> > >> > > > >> > > Once it is included, the cluster can be downgraded fine. > >> > > > >> > > I wonder how users would know that their existing deployment has > >> > > HBASE-16189 <https://issues.apache.org/jira/browse/HBASE-16189> > before > >> > > upgrading to 2.0 release. > >> > > > >> > > On Tue, Dec 6, 2016 at 2:29 AM, ramkrishna vasudevan < > >> > > [email protected]> wrote: > >> > > > >> > >> @Ted > >> > >> Does your version have this fix > >> > >> https://issues.apache.org/jira/browse/HBASE-16189 > >> > >> > >> > >> Regards > >> > >> Ram > >> > >> > >> > >> On Tue, Dec 6, 2016 at 3:56 PM, Ted Yu <[email protected]> > wrote: > >> > >> > >> > >> > Is the assumption that hbase:meta would not split ? > >> > >> > > >> > >> > In other thread, Francis Liu was proposing supporting splittable > >> > >> > hbase:meta in 2.0 release. > >> > >> > > >> > >> > Cheers > >> > >> > > >> > >> > > On Dec 6, 2016, at 2:20 AM, 张铎 <[email protected]> wrote: > >> > >> > > > >> > >> > > Could this be solved by hosting meta only on master? > >> > >> > > > >> > >> > > BTW, MetaCellComparator is introduced in HBASE-10800. > >> > >> > > > >> > >> > > Thanks. > >> > >> > > > >> > >> > > 2016-12-06 17:44 GMT+08:00 Ted Yu <[email protected]>: > >> > >> > > > >> > >> > >> Hi, > >> > >> > >> When I restarted a cluster with 1.1 , I found that hbase:meta > >> > region > >> > >> > >> (written to by the previously deployed 2.0) couldn't be > opened: > >> > >> > >> > >> > >> > >> Caused by: java.io.IOException: > >> > >> > >> org.apache.hadoop.hbase.io.hfile.CorruptHFileException: > Problem > >> > >> reading > >> > >> > >> HFile Trailer from file hdfs://yz1.xx.com:8020/apps/ > >> > >> hbase/data/data/ > >> > >> > >> hbase/meta/1588230740/info/599fc8a37311414e876803312009a986 > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore.openStoreFiles( > >> > >> HStore.java: > >> > >> > >> 579) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore.loadStoreFiles( > >> > >> HStore.java: > >> > >> > >> 534) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore.<init>( > >> > HStore.java:275) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.instantiateHSto > >> > >> re(HRegion. > >> > >> > >> java:5150) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HRegion$1.call(HRegion. > >> > >> java:912) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HRegion$1.call(HRegion. > >> > >> java:909) > >> > >> > >> at java.util.concurrent.FutureTask.run(FutureTask. > >> java:266) > >> > >> > >> at > >> > >> > >> java.util.concurrent.Executors$RunnableAdapter.call( > >> > >> Executors.java:511) > >> > >> > >> at java.util.concurrent.FutureTask.run(FutureTask. > >> java:266) > >> > >> > >> ... 3 more > >> > >> > >> Caused by: org.apache.hadoop.hbase.io. > >> hfile.CorruptHFileException: > >> > >> > Problem > >> > >> > >> reading HFile Trailer from file hdfs:// > >> > >> > >> yz1.xx.com:8020/apps/hbase/data/data/hbase/ meta/1588230740/ > >> > >> > >> info/599fc8a37311414e876803312009a986 > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion( > >> > >> > HFile.java:483) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.HFile.createReader( > >> > HFile.java:511) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.StoreFile$Reader. > >> > >> > >> <init>(StoreFile.java:1128) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.StoreFileInfo. > >> > >> > >> open(StoreFileInfo.java:267) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.StoreFile.open(StoreFil > >> > >> e.java:409) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.StoreFile. > >> > >> > >> createReader(StoreFile.java:517) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore.createStoreFileA > >> > >> ndReader( > >> > >> > >> HStore.java:687) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore.access$000( > >> > >> HStore.java:130) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore$1.call( > >> > HStore.java:554) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.regionserver.HStore$1.call( > >> > HStore.java:551) > >> > >> > >> ... 6 more > >> > >> > >> Caused by: java.io.IOException: java.lang. > >> ClassNotFoundException: > >> > >> > >> org.apache.hadoop.hbase.CellComparator$MetaCellComparator > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.FixedFileTrailer.getCompara > >> > >> torClass( > >> > >> > >> FixedFileTrailer.java:581) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.FixedFileTrailer. > >> > deserializeFromPB( > >> > >> > >> FixedFileTrailer.java:300) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.FixedFileTrailer. > >> > >> > >> deserialize(FixedFileTrailer.java:242) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.FixedFileTrailer. > >> readFromStream( > >> > >> > >> FixedFileTrailer.java:407) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion( > >> > >> > HFile.java:468) > >> > >> > >> ... 15 more > >> > >> > >> Caused by: java.lang.ClassNotFoundException: > >> > >> > >> org.apache.hadoop.hbase.CellComparator$MetaCellComparator > >> > >> > >> at java.net.URLClassLoader. > findClass(URLClassLoader.java: > >> > 381) > >> > >> > >> at java.lang.ClassLoader.loadClass(ClassLoader.java: > 424) > >> > >> > >> at sun.misc.Launcher$AppClassLoader.loadClass( > >> > Launcher.java: > >> > >> 331) > >> > >> > >> at java.lang.ClassLoader.loadClass(ClassLoader.java: > 357) > >> > >> > >> at java.lang.Class.forName0(Native Method) > >> > >> > >> at java.lang.Class.forName(Class.java:264) > >> > >> > >> at > >> > >> > >> org.apache.hadoop.hbase.io.hfile.FixedFileTrailer.getCompara > >> > >> torClass( > >> > >> > >> FixedFileTrailer.java:579) > >> > >> > >> > >> > >> > >> When user does rolling upgrade from 1.1 to 2.0, the above may > >> cause > >> > >> > problem > >> > >> > >> if hbase:meta region is updated by server running 2.0 but > later > >> > >> > assigned to > >> > >> > >> a region server which still runs 1.1 (due to crash of the > server > >> > >> running > >> > >> > >> 2.0, e.g.) > >> > >> > >> > >> > >> > >> I want to get community feedback on the severity of this > issue. > >> > >> > >> > >> > >> > >> Thanks > >> > >> > >> > >> > >> > > >> > >> > >> > > > >> > > > >> > > >> >
