Hey Jon, About table level TTL -> It wasn’t for optimization, just a suggestion. The user had no table level TTL set. It was at default 0. So if an insert comes in with no TTL, that row would never expire. There is no default TTL to fall back on in his case. Just thinking about possible problems with that setup and data never expiring.
About inc repair -> Yup, I agree, I did some extensive research into that earlier this year. We’re just waiting for 4.0 as well. But for some of our larger clusters, doing a full repair, even if it’s just with the –pr option is just not feasible. It takes weeks to repair and is a huge load on the cluster. So I’m continuing to use Inc repair and keeping an eye out for any issues. And I’m using –pr for the smaller clusters. But for tables with TWCS, I wouldn’t do a Full repair. Because it would include all the old and newer sstables in the operation and it is bound to cause new sstables with mixed data and may eventually lead to sstable blockers. It also kind of defeats the purpose of separating our data into neat TWCS windows. And therefore would take a while for tombstone cleanup too. Thank you, Meg From: Jonathan Haddad [mailto:j...@jonhaddad.com] Sent: Wednesday, October 24, 2018 1:42 PM To: user <user@cassandra.apache.org> Subject: Re: TWCS: Repair create new buckets with old data Hey Meg, a couple thoughts. > Set a table level TTL with TWCS, and stop setting it with inserts/updates > (insert TTL overrides table level TTL). So, that your entire sstable expires > at the same time, as opposed to each insert expiring at its own pace. So that > for tombstone clean up, the system can just drop the entire sstable at once. Setting the TTL on a table or the query gives you the same exact result. Setting it on the table is just there for convenience. If it's not set at the query level, it uses the default value. See org.apache.cassandra.cql3.Attributes#getTimeToLive. Generally speaking I'd rather set it at the table level as well, but it's just to avoid weird application bugs, not as an optimization. > I’d suggest removing the -pr. Running incremental repair with TWCS is > better. If incremental repair worked correctly I would agree with you, but unfortunately it doesn't. Incremental repair has some subtle bugs that can result in massive overstreaming due to the fact that it will sometimes not correctly mark data as repaired. My coworker Alex wrote up a good summary on the changes to incremental going into 4.0 to fix these issues, it's worth a read. http://thelastpickle.com/blog/2018/09/10/incremental-repair-improvements-in-cassandra-4.html. Reaper (OSS, maintained by us @ TLP, see http://cassandra-reaper.io/) has the ability to schedule subrange repairs on one or more tables, or all tables except those in a blacklist. Doing frequent subrange repairs will limit the amount of data that will get streamed in and should help keep things pretty consistent unless you're dropping a lot of mutations. It's not perfect but should cause less headache than incremental repair will. Hope this helps. Jon On Thu, Oct 25, 2018 at 4:21 AM Meg Mara <mm...@digitalriver.com<mailto:mm...@digitalriver.com>> wrote: Hi Maik, I have a similar Cassandra env, with similar table requirements. So these would be my suggestions: • Set a table level TTL with TWCS, and stop setting it with inserts/updates (insert TTL overrides table level TTL). So, that your entire sstable expires at the same time, as opposed to each insert expiring at its own pace. So that for tombstone clean up, the system can just drop the entire sstable at once. • Since you’re on v3.0.9, nodetool repair command runs incremental repair by default. And with inc repair, -pr option is not recommended. (ref. link below) • I’d suggest removing the -pr. Running incremental repair with TWCS is better. • Here’s why I think so -> Full repair and Full repair with –PR option would include all the sstables in the repair process, which means the chance of your oldest and newest data mixing is very high. • Whereas, if you run incremental repair every 5 days for example, only the last five days of data would be included in that repair operation. So, the maximum ‘damage’ it would do is mixing 5 day old data in a new sstable. • Your table level TTL would then tombstone this data on 4 month + 5 day mark instead of on the 4 month mark. Which shouldn’t be a big concern. At least in our case it isn’t! • I wouldn’t stop running repairs on our TWCS tables, because we are too concerned with data consistency. Please read the note here: https://docs.datastax.com/en/cassandra/3.0/cassandra/tools/toolsRepair.html Thank you, Meg From: Caesar, Maik [mailto:maik.cae...@dxc.com<mailto:maik.cae...@dxc.com>] Sent: Wednesday, October 24, 2018 2:17 AM To: user@cassandra.apache.org<mailto:user@cassandra.apache.org> Subject: RE: TWCS: Repair create new buckets with old data Hi Meg, the ttl (4 month) is set during insert via insert statement with the application. The repair is started each day on one of ten hosts with command : nodetool --host hostname_# repair –pr Regards Maik From: Meg Mara [mailto:mm...@digitalriver.com] Sent: Dienstag, 23. Oktober 2018 17:05 To: user@cassandra.apache.org<mailto:user@cassandra.apache.org> Subject: RE: TWCS: Repair create new buckets with old data Hi Maik, I noticed in your table description that your default_time_to_live = 0, so where is your TTL property set? At what point do your sstables get tombstoned? Also, could you please mention what kind of repair you performed on this table? (Incremental, Full, Full repair with -pr option) Thank you, Meg From: Caesar, Maik [mailto:maik.cae...@dxc.com] Sent: Monday, October 22, 2018 10:17 AM To: user@cassandra.apache.org<mailto:user@cassandra.apache.org> Subject: RE: TWCS: Repair create new buckets with old data Ok, thanks. My conclusion: 1. I will set unchecked_tombstone_compaction to true to get old data with tombstones removed 2. I will exclude TWCS tables from repair Regarding exclude table from repair, is there any easy way to do this? Nodetool repaire do not support excludes. Regards Maik From: wxn...@zjqunshuo.com<mailto:wxn...@zjqunshuo.com> [mailto:wxn...@zjqunshuo.com] Sent: Freitag, 19. Oktober 2018 03:58 To: user <user@cassandra.apache.org<mailto:user@cassandra.apache.org>> Subject: RE: TWCS: Repair create new buckets with old data > Is the repair not necessary to get data files remove from filesystem ? The answer is no. IMO, Cassandra will remove sstable files automatically if it can make sure the sstable files are 100% of tombstones and safe to do deletion. If you use TWCS and you have only insertion and no update, you don't need run repair manually. -Simon From: Caesar, Maik<mailto:maik.cae...@dxc.com> Date: 2018-10-18 20:30 To: user@cassandra.apache.org<mailto:user@cassandra.apache.org> Subject: RE: TWCS: Repair create new buckets with old data Hello Simon, Is the repair not necessary to get data files remove from filesystem ? My assumption was, that only repaired data will removed after TTL is reached. Regards Maik From: wxn...@zjqunshuo.com<mailto:wxn...@zjqunshuo.com> [mailto:wxn...@zjqunshuo.com] Sent: Mittwoch, 17. Oktober 2018 09:02 To: user <user@cassandra.apache.org<mailto:user@cassandra.apache.org>> Subject: Re: TWCS: Repair create new buckets with old data Hi Maik, IMO, when using TWCS, you had better not run repair. The behaviour of TWCS is same with STCS for repair when merging sstables and the result is leaving sstables spanning multiple time buckets, but maybe I'm wrong. In my use case, I don't do repair with table using TWCS. -Simon From: Caesar, Maik<mailto:maik.cae...@dxc.com> Date: 2018-10-16 17:46 To: user@cassandra.apache.org<mailto:user@cassandra.apache.org> Subject: TWCS: Repair create new buckets with old data Hallo, we work with Cassandra version 3.0.9 and have a problem in a table with TWCS. The command “nodetool repair” create always new files with old data. This avoid the delete of the old data. The layout of the Table is following: cqlsh> desc stat.spa CREATE TABLE stat.spa ( region int, id int, date text, hour int, zippedjsonstring blob, PRIMARY KEY ((region, id), date, hour) ) WITH CLUSTERING ORDER BY (date ASC, hour ASC) AND bloom_filter_fp_chance = 0.01 AND caching = {'keys': 'ALL', 'rows_per_partition': 'NONE'} AND comment = '' AND compaction = {'class': 'org.apache.cassandra.db.compaction.TimeWindowCompactionStrategy', 'compaction_window_size': '1', 'compaction_window_unit': 'DAYS', 'max_threshold': '100', 'min_threshold': '4', 'tombstone_compaction_interval': '86460'} AND compression = {'chunk_length_in_kb': '64', 'class': 'org.apache.cassandra.io.compress.LZ4Compressor'} AND crc_check_chance = 1.0 AND dclocal_read_repair_chance = 0.0 AND default_time_to_live = 0 AND gc_grace_seconds = 864000 AND max_index_interval = 2048 AND memtable_flush_period_in_ms = 0 AND min_index_interval = 128 AND read_repair_chance = 0.0 AND speculative_retry = '99PERCENTILE'; Actual the oldest data are from 2017/04/15 and will not remove: $ for f in *Data.db; do meta=$(sudo sstablemetadata $f); echo -e "Max:" $(date --date=@$(echo "$meta" | grep Maximum\ time | cut -d" " -f3| cut -c 1-10) '+%Y/%m/%d %H:%M') "Min:" $(date --date=@$(echo "$meta" | grep Minimum\ time | cut -d" " -f3| cut -c 1-10) '+%Y/%m/%d %H:%M') $(echo "$meta" | grep droppable) $(echo "$meta" | grep "Repaired at") ' \t ' $(ls -lh $f | awk '{print $5" "$6" "$7" "$8" "$9}'); done | sort Max: 2017/04/15 12:08 Min: 2017/03/31 13:09 Estimated droppable tombstones: 1.7731048805815162 Repaired at: 1525685601400 42K May 7 19:56 mc-22922-big-Data.db Max: 2017/04/17 13:49 Min: 2017/03/31 13:09 Estimated droppable tombstones: 1.9600207684319835 Repaired at: 1525685601400 116M May 7 13:31 mc-15096-big-Data.db Max: 2017/04/21 13:43 Min: 2017/04/15 13:34 Estimated droppable tombstones: 1.9090909090909092 Repaired at: 1525685601400 11K May 7 19:56 mc-22921-big-Data.db Max: 2017/05/23 21:45 Min: 2017/04/21 14:00 Estimated droppable tombstones: 1.8360655737704918 Repaired at: 1525685601400 21M May 7 19:56 mc-22919-big-Data.db Max: 2017/06/12 15:19 Min: 2017/04/25 14:45 Estimated droppable tombstones: 1.8091397849462365 Repaired at: 1525685601400 19M May 7 14:36 mc-17095-big-Data.db Max: 2017/06/15 15:26 Min: 2017/05/10 14:37 Estimated droppable tombstones: 1.76536312849162 Repaired at: 1529612605539 9.3M Jun 21 22:31 mc-25372-big-Data.db … After a „nodetool repair“ run, a new big data file is created that include old data from 2017/07/31. Max: 2018/07/27 18:10 Min: 2017/03/31 13:13 Estimated droppable tombstones: 0.08392555471691247 Repaired at: 0 11G Sep 11 22:02 mc-39281-big-Data.db … Max: 2018/08/16 18:18 Min: 2018/08/06 12:19 Estimated droppable tombstones: 0.0 Repaired at: 1534525730510 123M Aug 17 23:46 mc-36847-big-Data.db Max: 2018/08/17 19:20 Min: 2017/07/31 12:04 Estimated droppable tombstones: 0.03385963490004347 Repaired at: 0 11G Sep 11 21:43 mc-39265-big-Data.db Max: 2018/08/17 19:20 Min: 2018/07/24 12:33 Estimated droppable tombstones: 0.0 Repaired at: 1534525730510 135M Sep 11 21:44 mc-39270-big-Data.db … Max: 2018/09/06 17:30 Min: 2018/08/28 12:17 Estimated droppable tombstones: 0.0 Repaired at: 1536690786879 129M Sep 11 21:10 mc-39238-big-Data.db Max: 2018/09/07 18:22 Min: 2017/04/23 12:48 Estimated droppable tombstones: 0.1548442441468401 Repaired at: 0 8.0G Sep 11 21:33 mc-39258-big-Data.db Max: 2018/09/07 18:22 Min: 2018/09/07 12:15 Estimated droppable tombstones: 0.0 Repaired at: 1536690786879 72M Sep 11 21:34 mc-39262-big-Data.db Max: 2018/09/08 18:20 Min: 2018/08/22 12:17 Estimated droppable tombstones: 0.0 Repaired at: 0 2.8G Sep 11 21:47 mc-39272-big-Data.db The tool sstableexpiredblockers shows that the file mc-39281-big-Data.db blocks 95 expired files from getting dropped, for example the oldest file mc-22922-big-Data.db [BigTableReader(path='.../stat/spa-.../mc-39281-big-Data.db') (minTS = 1490958782530000, maxTS = 1532707837676719, maxLDT = 1557154990) blocks 95 expired sstables from getting dropped: [BigTableReader(path='.../stat/spa-.../mc-36936-big-Data.db') (minTS = 1500027128958000, maxTS = 1503666765807229, maxLDT = 1535202765) [BigTableReader(path='.../stat/spa-.../mc-22921-big-Data.db') (minTS = 1492256093314000, maxTS = 1492775013454001, maxLDT = 1524311013) [BigTableReader(path='.../stat/spa-.../mc-36947-big-Data.db') (minTS = 1492255708403000, maxTS = 1501937182477001, maxLDT = 1533473182) [BigTableReader(path='.../stat/spa-.../mc-32582-big-Data.db') (minTS = 1493028031639000, maxTS = 1499175057476001, maxLDT = 1530711057) [BigTableReader(path='.../stat/spa-.../mc-32560-big-Data.db') (minTS = 1500210297826000, maxTS = 1501416691390001, maxLDT = 1532952691) [BigTableReader(path='.../stat/spa-.../mc-32528-big-Data.db') (minTS = 1490958761762000, maxTS = 1504358072394248, maxLDT = 1535894072) [BigTableReader(path='.../stat/spa-.../mc-32572-big-Data.db') (minTS = 1500027103795000, maxTS = 1500297137808001, maxLDT = 1531833137) [BigTableReader(path='.../stat/spa-.../mc-36935-big-Data.db') (minTS = 1500038582669000, maxTS = 1503839159485824, maxLDT = 1535375159) [BigTableReader(path='.../stat/spa-.../mc-22922-big-Data.db') (minTS = 1490958570018000, maxTS = 1492250905633001, maxLDT = 1523786905) [BigTableReader(path='.../stat/spa-.../mc-33470-big-Data.db') (minTS = 1499940836241000, maxTS = 1500040376685000, maxLDT = 1531576376) Why create the repair such turbulence in new data files and how can we remove the old data? Kind Regards Maik Cäsar DXC Technology Company -- This message is transmitted to you by or on behalf of DXC Technology Company or one of its affiliates. It is intended exclusively for the addressee. The substance of this message, along with any attachments, may contain proprietary, confidential or privileged information or information that is otherwise legally exempt from disclosure. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient of this message, you are not authorized to read, print, retain, copy or disseminate any part of this message. If you have received this message in error, please destroy and delete all copies and notify the sender by return e-mail. Regardless of content, this e-mail shall not operate to bind DXC Technology Company or any of its affiliates to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose. --. -- Jon Haddad http://www.rustyrazorblade.com twitter: rustyrazorblade
