Re: March 2015 QA retrospective
CASSANDRA-8285 https://issues.apache.org/jira/browse/CASSANDRA-8285 Aleksey Yeschenko Move all hints related tasks to hints private executor Pierre's reproducer represents something we weren't doing, but that users are. Is that now being tested? That particular issue will not happen again. That class of issues can only be tested by a sufficiently long running stress test, with plenty of chaos-monkeying thrown in. That's essentially how it got caught - by driver duration tests with chaos-monkeying in. It's still being run exercised by the driver tests, and we do now run them prior to releasing stuff, so I'd say yes - it's being tested. Once we have our own framework, we'll migrate it there, from driver tests. CASSANDRA-8462 https://issues.apache.org/jira/browse/CASSANDRA-8462 Aleksey Yeschenko Upgrading a 2.0 to 2.1 breaks CFMetaData on 2.0 nodes Have additional dtest coverage, need to do this in kitchen sink tests dtests coverage would not help here, not really. It's a tricky race condition that can only be triggered during upgrade. Would need some kind of test that repeatedly upgrades the nodes in the cluster, again and again, to have this caught, or else it'd be flaky at best. The actual proper fix for that issue is the new, versioned, schema update exchange protocol - https://issues.apache.org/jira/browse/CASSANDRA-6038 is the ticket. That one will come with a metric ton of tests. On Thu, Apr 9, 2015 at 11:45 AM, Ariel Weisberg ariel.weisb...@datastax.com wrote: Repeated with sort *Key* *Assignee* *Summary* *Revisit reason* CASSANDRA-8285 https://issues.apache.org/jira/browse/CASSANDRA-8285 Aleksey Yeschenko Move all hints related tasks to hints private executor Pierre's reproducer represents something we weren't doing, but that users are. Is that now being tested? CASSANDRA-8462 https://issues.apache.org/jira/browse/CASSANDRA-8462 Aleksey Yeschenko Upgrading a 2.0 to 2.1 breaks CFMetaData on 2.0 nodes Have additional dtest coverage, need to do this in kitchen sink tests CASSANDRA-8640 https://issues.apache.org/jira/browse/CASSANDRA-8640 Anthony Cozzie Paxos requires all nodes for CAS If PAXOS is not supposed to require all nodes for CAS we should be able to fail nodes or a certain number of nodes and still continue to CAS (test availability of CAS under failure conditions). No regression test. CASSANDRA-8677 https://issues.apache.org/jira/browse/CASSANDRA-8677 Ariel Weisberg rpc_interface and listen_interface generate NPE on startup when specified interface doesn't exist Missing unit tests checking error messages for DatabaseDescriptor CASSANDRA-8577 https://issues.apache.org/jira/browse/CASSANDRA-8577 Artem Aliev Values of set types not loading correctly into Pig Full set of interactions with PIG not validated CASSANDRA-7704 https://issues.apache.org/jira/browse/CASSANDRA-7704 Benedict FileNotFoundException during STREAM-OUT triggers 100% CPU usage Streaming testing didn't reproduce this before release CASSANDRA-8383 https://issues.apache.org/jira/browse/CASSANDRA-8383 Benedict Memtable flush may expire records from the commit log that are in a later memtable No regression test, no follow up ticket. Could/should this have been reproducable as an actual bug? CASSANDRA-8429 https://issues.apache.org/jira/browse/CASSANDRA-8429 Benedict Some keys unreadable during compaction Running stress in CI would have caught this, and we're going to do that CASSANDRA-8459 https://issues.apache.org/jira/browse/CASSANDRA-8459 Benedict autocompaction on reads can prevent memtable space reclaimation What would have reproduced this before release? CASSANDRA-8499 https://issues.apache.org/jira/browse/CASSANDRA-8499 Benedict Ensure SSTableWriter cleans up properly after failure Testing error paths? Any way to test things in a loop to detect leaks? CASSANDRA-8513 https://issues.apache.org/jira/browse/CASSANDRA-8513 Benedict SSTableScanner may not acquire reference, but will still release it when closed This had a user visible component, what test could have caught it befor erelease? CASSANDRA-8619 https://issues.apache.org/jira/browse/CASSANDRA-8619 Benedict using CQLSSTableWriter gives ConcurrentModificationException What kind of test would have caught this before release? CASSANDRA-8632 https://issues.apache.org/jira/browse/CASSANDRA-8632 Benedict cassandra-stress only generating a single unique row We rely on stress for performance testing, that might mean it needs real testing that demonstrates it generates load that looks like the load it is supposed to be generating. CASSANDRA-8668 https://issues.apache.org/jira/browse/CASSANDRA-8668 Benedict We don't enforce offheap memory constraints; regression introduced by 7882 Memory constraints was a supported feature/UI, but not completely tested before release. Could this have been found most effectively by a unit test or a blackbox test? CASSANDRA-8719
Re: [discuss] Modernization of Cassandra build system
Hey Benedict, My replies in line According to some recordings from DataStax there is a plan to support in Cassandra multiple kinds of store - document, graph so it won’t get easier with the time but rather harder - ask yourself do you really want to mess all these things together? Well, these certainly won't live in the same repository, so I wouldn't worry about that That’s good. That’s very good cause it will force separation. If you will do that please consider using other build system to don’t repeat mistakes which are present now in main Cassandra build. As I briefly counted in my ealier mail there was 116 issues related to artifacts published by build process. That does sound like a lot of bugs. How many actual maintenance releases were necessary, did you happen to also count? This is something that could be raised at the new retrospective that Ariel has begun, to see if there's anything that can be done to reduce their incidence and risk. There have been 159 minor releases of cassandra (git tag —list | egrep rc | egrep beta | wc -l). I did not track exactly what is correnation of the bug ration. These 116 vs 159 are just numbers. From my understanding there is 116 unecessary issues which could be avoided. You can read these numbers in two different ways - every second minor release was fixing maven artifacts OR every second release was broken due the maven artifacts. Seems you preffer first one while users usualy observes second. however it gives real boost when it comes to community donations, tool development, or even debugging You're conflating the task of upgrading the build system with modularisation, which is a bad idea if you want to make progress on either one, since they're each a different and difficult discussion, even if they relate. I do that cause this is typical chicken vs egg problem. One thing can not be done without another it’s just question which one is fist to follow. Code modularization/package separation without strict bounds is hard to follow. However nothing prevents doing this in reverse mode - by solving code issues first and then introducing new build tool. It’s up to cassandra developers to decide. On the topic of the build system: if you can justify why you think Maven has a significant chance of reducing our bug burden here, a case can perhaps be made, and I will defer to the members of this list with more experience of our build system for that in depth discussion. At the moment, it seems to be taken as a given this would occur, but I don't yet see a clear reason that we should expect this to occur. You see - I don’t have to justify Maven. I have proposed you a help with it. I also gave you couple of reasons why Ant is not first sort of tools these days. I don’t feel myself responsible for doing any advocating for Maven itself. It’s up to you what you choose. The major thing, major problem which modern tools are doing for you is build time classpath management (both compile test) and separate javac executions for both of these. Take what you preffer - gradle, sbt, leiningen. Anything which does things from previous sentence. Do your own evaluation. Take what work for you, not only for me. On the topic of modularisation: Like I said previously, everyone on this list is sympathetic to that goal, I think. However the practical reality is likely to be too confounding. But that doesn't mean it is absolutely a losing battle, if you can demonstrate a sufficiently painless and worthwhile transition. I don't quite get you at this point. From one side you suppose everyone is for taking such step, from another one you ask for proofs. In case of code relocation there are always multiple ways. Cause of what you have currently forces solution of multiple problems. You can start on any of it (ie. circular dependencies I did mention in earlier conversation doesn’t require changing a tool). In place where you stay at this moment there will be no such thing as painless transition. As said ealier - it will be only harder over time. Given example from my life. We do use Cassandra. We do have plenty of mid level integration tests which are verifying end to end functionality. Starting from frontend or messaging layer up to data persistence. Now each of our tests even if it consist a low amount of data hits IO on multiple levels - starting from socket ending on disk. We do not test in such cases consistency levels as it’s assumed to be tested by cassandra itself - we are ensuring that incoming data passes storage interface and can be retrieved back via same interface. With what cassandra is now we can not make our tests running fast. People are prisoners of cassandra-unit cause embedding cassandra is impossible, even if it’s written using portable language. It has too many inner and outer dependencies. On other hand we have for example ActiveMQ which has lots of options. Even with all of these it might be embedded
Re: 3.0 and the Cassandra release process
On Tue, Mar 17, 2015 at 4:06 PM, Jonathan Ellis jbel...@gmail.com wrote: I’m optimistic that as we improve our process this way, our even releases will become increasingly stable. If so, we can skip sub-minor releases (3.2.x) entirely, and focus on keeping the release train moving. In the meantime, we will continue delivering 2.1.x stability releases. The weak point of this plan is the transition from the big release development methodology culminating in 3.0, to the monthly tick-tock releases. Since 3.0 needs to go through a beta/release candidate phase, during which we're going to be serious about not adding new features, that means that 3.1 will come with multiple months worth of features, so right off the bat we're starting from a disadvantage from a stability standpoint. Recognizing that it will take several months for the tick-tock releases to stabilize, I would like to ship 3.0.x stability releases concurrently with 3.y tick-tock releases. This should stabilize 3.0.x faster than tick-tock, while at the same time hedging our bets such that if we assess tick-tock in six months and decide it's not delivering on its goals, we're not six months behind in having a usable set of features that we shipped in 3.0. So, to summarize: - New features will *only* go into tick-tock releases. - Bug fixes will go into tick-tock releases and a 3.0.x branch, which will be maintained for at least a year -- Jonathan Ellis Project Chair, Apache Cassandra co-founder, http://www.datastax.com @spyced
Impact of removing compactions_in_progress folder
Often we face errors on Cassandra start regarding unfinished compactions particularly when cassandra was abrupty shut down . Problem gets resolved when we delete /var/lib/cassandra/data/system/compactions_in_progress folder. Does deletion of the folder has any impact on integrity of data or any other aspect? Thanks Anuj Wadehra
Drawbacks of Major Compaction now that Automatic Tombstone Compaction Exists
Recently we faced an issue where every repair operation caused addition of hundreds of sstables (CASSANDRA-9146). In order to bring situation under control and make sure reads are not impacted, we were left with no option but to run major compaction to ensure that thousands of tiny sstables are compacted. Queries: Does major compaction has any drawback after automatic tombstone compaction got implemented in 1.2 via tombstone_threshold sub-property(CASSANDRA-3442)? I understand that the huge SSTable created after major compaction wont be compacted with new data any time soon but is that a problem if purged data is removed via automatic tombstone compaction? If we major compaction results in a huge file say 500GB, what are the drawbacks of it? If one big sstable is a problem, is there any way of solving the problem? We tried running sstablesplit after major compaction to split the big sstable but as new sstables were of same size they are again compacted into single huge table once Cassandra was started after executing sstablesplit. Thanks Anuj Wadehra
Re: Drawbacks of Major Compaction now that Automatic Tombstone Compaction Exists
I havent got much response regarding this on user list..so posting it on dev list too.. Thanks Anuj Wadehra Sent from Yahoo Mail on Android From:Anuj Wadehra anujw_2...@yahoo.co.in Date:Tue, 14 Apr, 2015 at 7:05 am Subject:Drawbacks of Major Compaction now that Automatic Tombstone Compaction Exists Recently we faced an issue where every repair operation caused addition of hundreds of sstables (CASSANDRA-9146). In order to bring situation under control and make sure reads are not impacted, we were left with no option but to run major compaction to ensure that thousands of tiny sstables are compacted. Queries: Does major compaction has any drawback after automatic tombstone compaction got implemented in 1.2 via tombstone_threshold sub-property(CASSANDRA-3442)? I understand that the huge SSTable created after major compaction wont be compacted with new data any time soon but is that a problem if purged data is removed via automatic tombstone compaction? If we major compaction results in a huge file say 500GB, what are the drawbacks of it? If one big sstable is a problem, is there any way of solving the problem? We tried running sstablesplit after major compaction to split the big sstable but as new sstables were of same size they are again compacted into single huge table once Cassandra was started after executing sstablesplit. Thanks Anuj Wadehra
Re: [discuss] Modernization of Cassandra build system
every second minor release was fixing maven artifacts OR every second release was broken due the maven artifacts Well, it's also possible just one release had 116 build artefact problems? Obviously that's the absurd extreme end, but the reason I was asking if you had any idea, since you'd done the counting. I don’t feel myself responsible for doing any advocating for Maven itself. It’s up to you what you choose. This is a community process, and I'm trying (and apparently failing) to help you understand at least how *I* understand it to work, and the problems I see with what you're proposing. The silence on the list suggests there is significant inertia and no other strong advocates for this change. This could be for myriad reasons, from people simply not caring, to thinking there are roughly equal pros and cons, to also just hoping the conversation will go away because they're against it. Without advocacy, the inertia is not overcome, and since you're the only person so far to express a desire for this change, it is unfortunately up to you to convince us. I, and I'm sure the rest of the community, are very appreciative of the offer of your time. We really are. Unfortunately that isn't enough to warrant utilising it, but we *are* open to discussion and advocacy on the topic. The crux of the problem is that Cassandra has a lot of important work being done to it, work that I personally perceive (and suspect others do also) as more important than the admitted inadequacy of our modularisation and, perhaps, our build system (I plead ignorance here). This work is currently surpassing the labour we have to address it. If this upheaval hinders that work, that is bad, and that is what I mean when I say warrants - is the upheaval small enough, or the yield really great (modularisation doesn't always pan out, so we may not even get a good result, but still have the significant pain)? I don't want to give you the impression I am either a gatekeeper or shooting down your proposal. I'm just attempting to explain my perception of the view of the existing contributors. On Mon, Apr 13, 2015 at 9:31 PM, Łukasz Dywicki l...@code-house.org wrote: Hey Benedict, My replies in line According to some recordings from DataStax there is a plan to support in Cassandra multiple kinds of store - document, graph so it won’t get easier with the time but rather harder - ask yourself do you really want to mess all these things together? Well, these certainly won't live in the same repository, so I wouldn't worry about that That’s good. That’s very good cause it will force separation. If you will do that please consider using other build system to don’t repeat mistakes which are present now in main Cassandra build. As I briefly counted in my ealier mail there was 116 issues related to artifacts published by build process. That does sound like a lot of bugs. How many actual maintenance releases were necessary, did you happen to also count? This is something that could be raised at the new retrospective that Ariel has begun, to see if there's anything that can be done to reduce their incidence and risk. There have been 159 minor releases of cassandra (git tag —list | egrep rc | egrep beta | wc -l). I did not track exactly what is correnation of the bug ration. These 116 vs 159 are just numbers. From my understanding there is 116 unecessary issues which could be avoided. You can read these numbers in two different ways - every second minor release was fixing maven artifacts OR every second release was broken due the maven artifacts. Seems you preffer first one while users usualy observes second. however it gives real boost when it comes to community donations, tool development, or even debugging You're conflating the task of upgrading the build system with modularisation, which is a bad idea if you want to make progress on either one, since they're each a different and difficult discussion, even if they relate. I do that cause this is typical chicken vs egg problem. One thing can not be done without another it’s just question which one is fist to follow. Code modularization/package separation without strict bounds is hard to follow. However nothing prevents doing this in reverse mode - by solving code issues first and then introducing new build tool. It’s up to cassandra developers to decide. On the topic of the build system: if you can justify why you think Maven has a significant chance of reducing our bug burden here, a case can perhaps be made, and I will defer to the members of this list with more experience of our build system for that in depth discussion. At the moment, it seems to be taken as a given this would occur, but I don't yet see a clear reason that we should expect this to occur. You see - I don’t have to justify Maven. I have proposed you a help with it. I also gave you couple of reasons why Ant is not first sort of tools these
Re: [discuss] Modernization of Cassandra build system
According to some recordings from DataStax there is a plan to support in Cassandra multiple kinds of store - document, graph so it won’t get easier with the time but rather harder - ask yourself do you really want to mess all these things together? Well, these certainly won't live in the same repository, so I wouldn't worry about that As I briefly counted in my ealier mail there was 116 issues related to artifacts published by build process. That does sound like a lot of bugs. How many actual maintenance releases were necessary, did you happen to also count? This is something that could be raised at the new retrospective that Ariel has begun, to see if there's anything that can be done to reduce their incidence and risk. however it gives real boost when it comes to community donations, tool development, or even debugging You're conflating the task of upgrading the build system with modularisation, which is a bad idea if you want to make progress on either one, since they're each a different and difficult discussion, even if they relate. On the topic of the build system: if you can justify why you think Maven has a significant chance of reducing our bug burden here, a case can perhaps be made, and I will defer to the members of this list with more experience of our build system for that in depth discussion. At the moment, it seems to be taken as a given this would occur, but I don't yet see a clear reason that we should expect this to occur. On the topic of modularisation: Like I said previously, everyone on this list is sympathetic to that goal, I think. However the practical reality is likely to be too confounding. But that doesn't mean it is absolutely a losing battle, if you can demonstrate a sufficiently painless and worthwhile transition. On Sat, Apr 11, 2015 at 11:12 AM, Łukasz Dywicki l...@code-house.org wrote: Sorry for not coming back to topic for long time. You are right that what Cassandra project have currently - does work and keeping package scoping discipline in such big development community as Cassandra is clearly impossible without tool support (if you insist to keep ant please try to separate javac tasks for logical parts in current build to verify that). I clearly pointed out that it doesn’t work in reliable way causing troubles with artifacts uploaded to maven central. As I briefly counted in my ealier mail there was 116 issues related to artifacts published by build process. It is a lot and these changes requires another mainanance releases to fix for example one or another bytecode level dependency causing NoClassDefErrors with invalid artifacts. According to some recordings from DataStax there is a plan to support in Cassandra multiple kinds of store - document, graph so it won’t get easier with the time but rather harder - ask yourself do you really want to mess all these things together? Starting from 2.x Cassandra supports triggers but writing even a simplest trigger which will drop a log message or publish UDP packet requires entire cassandra and all it’s dependencies to be present during development. Fact that everything sits in one big ant build.xml is caused by troubles generated by ant itself to support multiple build modules, placeholders and so on, not because it’s handsome to do such. Modernization of build and internal dependencies is not something which brings huge benefit in first run cause now your frontend is CQL, however it gives real boost when it comes to community donations, tool development, or even debugging. Sadly keeping current Ant build is silent agreement to keep mess internally and rickety architecture of project. Ant was already legacy tool when Cassandra has been launched. The longer you will stay with it the more troubles you will get with it over time. Kind regards, Lukasz Wiadomość napisana przez Robert Stupp sn...@snazy.de w dniu 2 kwi 2015, o godz. 14:51: TL;DR - Benedict is right. IMO Maven is a nice, straight-forward tool if you know what you’re doing and start on a _new_ project. But Maven easily becomes a pita if you want to do something that’s not supported out-of-the-box. I bet that Maven would just not work for C* source tree with all the little nice features that C*’s build.xml offers (just look at the scripted stuff in build.xml). Eventually gradle would be an option; I proposed to switch to gradle several months ago. Same story (although gradle is better than Maven ;) ). But… you need to know that build.xml is not just used to build the code and artifacts. It is also used in CI, ccm, cstar-perf and a some other custom systems that exist and just work. So - if we would exchange ant with something else, it would force a lot of effort to change several tools and systems. And there must be a guarantee that everything works like it did before. Regarding IDEs: i’m using IDEA every day and it works like a charm with C*. Eclipse is ”supported natively” by
Re: March 2015 QA retrospective
On Fri, Apr 10, 2015 at 8:34 PM, Ariel Weisberg ariel.weisb...@datastax.com wrote: Hi Marcus, CASSANDRA-8211 https://issues.apache.org/jira/browse/CASSANDRA-8211 Overlapping sstables in L1+ So the question I would ask is would the workload that reproduces this specific bug be interesting in the general sense. Do we have to do anything special to reproduce it like make many non-verlapping sstables in L0 and is that interesting in the general sense. The way I reproduced it was (and I doubt it is interesting in the general case) * insert 10M keys with stress * major compact * stop node * sstablesplit * start node * switch to LCS * wait, repeat if no error within 5 minutes ie, very specific to the bug CASSANDRA-8320 https://issues.apache.org/jira/browse/CASSANDRA-8320 2.1.2: NullPointerException in SSTableWriter What tests did we add for this? Is it a targeted regression test, or are we now fully exercising SSTableRewriter the way users do against representative configurations and data? Yes, we added unit tests that exercises the code in SSTableRewriter CASSANDRA-8432 https://issues.apache.org/jira/browse/CASSANDRA-8432 Standalone Scrubber broken for LCS OK great. Looks like I didn't need to ask Jeremiah to do to look at the offline tools. CASSANDRA-8386 https://issues.apache.org/jira/browse/CASSANDRA-8386 Make sure we release references to sstables after incremental repair CASSANDRA-8316 https://issues.apache.org/jira/browse/CASSANDRA-8316 Did not get positive replies from all endpoints error on incremental repair CASSANDRA-8580 https://issues.apache.org/jira/browse/CASSANDRA-8580 AssertionErrors after activating unchecked_tombstone_compaction with leveled compaction CASSANDRA-8458 https://issues.apache.org/jira/browse/CASSANDRA-8458 Don't give out positions in an sstable beyond its first/last tokens Can you capture the elements of what we need and put it under CASSANDRA-9012. Maybe a ticket for testing repair under load, a ticket for switching compaction strategies and what the test scenarios for that would be, and when you say configurations what kind of configurations are you thinking of? will add subtickets to #9012 CASSANDRA-8525 https://issues.apache.org/jira/browse/CASSANDRA-8525 Bloom Filter truePositive counter not updated on key cache hit I think for JMX we only need to test that the access path delivers the value. Knowing we had to test for the value was I think an original implementer and reviewer issue. There was a metric exposed by the unit so we needed a test that shows the metric has a correct value. CASSANDRA-8532 https://issues.apache.org/jira/browse/CASSANDRA-8532 Fix calculation of expected write size during compaction https://issues.apache.org/jira/browse/CASSANDRA-8562 Fix checking available disk space before compaction starts Great. Linked 9154 to 9012 (Triage missing tests) CASSANDRA-8635 https://issues.apache.org/jira/browse/CASSANDRA-8635 STCS cold sstable omission does not handle overwrites without reads Can you create a ticket for this then? We should emit this workload pattern and then validate after that utilization is as expected. Alternatively this could be caught as a performance regression? It's starting to seem like some regressions could be caught as performance regressions. Yes this would definitely be caught in a performance test, if the workload was correct (ie, very few reads, many writes) /Marcus
Re: March 2015 QA retrospective
Hi Benedict, This only requires unit testing or dtests to be run this way. However for the kitchen sink tests this is just another dimension in the configuration state space, which IMO should be addressed as a whole methodically. Perhaps we should file a central JIRA, or the Google doc you suggested, for tracking all of these data points? I created a doc https://docs.google.com/a/datastax.com/document/d/1kccPqxEAoYQpT0gXnp20MYQUDmjOrakAeQhf6vkqjGo/edit?usp=sharing that is requirements, but not implementation. I want to list things we would like it to test in the general sense, as well as enumerating specific bugs that it should have been able to catch. This does raise an interesting, but probably not significant downside to the new approach: I fixed this ticket because somebody mentioned to me that it was hurting them, and I saw a quick and easy fix. The testing would not be quick and easy, so I am unlikely to volunteer to patch quick fixes in the new world order. This will certainly lead to higher quality bug fixes, but it may lead to fewer of them, and fewer instances of volunteer work to help people out, because the overhead eats too much into the work you're actually responsible for. This may lead to bug fixing being seen as much more of a chore than it already can be. I don't say this to discourage the new approach; it is just a thought that occurs to me off the back of this specific discussion. It's a real problem. People doing bugs fixes can be stuck spending months doing nothing but that and writing tests to fill in coverage. Then they get unhappy and unproductive. One of the reasons I leave the option for filing a JIRA open instead of saying that they have to do something is that it gives assignees and reviewers the option to have the work done later or by someone else. The person who is scheduling releases can see that the test issues before release (you would set fix version for the next release). It's still not done and the release is not done. That puts pressure on the person who wants to release to make sure it is in someone's queue. If you are hardcore agile and doing one or two week sprints what happens is that there are no tickets left in the sprint other than what was agreed on at at the planning meeting and people will have no choice but to work on test tasks. How we manage and prioritize tasks right now is magic to me and maybe not something that scales down to monthly releases. For monthly releases on at least a weekly basis you need to know what stands between you and the release being done and you need to have a plan for who is going to take care of the blockers that crop up. The testing would not be quick and easy, so I am unlikely to volunteer to patch quick fixes in the new world order. I think this gets into how we load balance bug fixes. There is a clear benefit to routing the bug to the person who will know how to fix and test it. I have never seen bugs as something you volunteer for. They typically belong somewhere and if it is with you then so be it. because the overhead eats too much into the work you're actually responsible for. We need to make sure that bug fixing isn't seen that way. I think it's important to make sure bugs find their way home. The work your actually responsible for is not done so you can't claim that bug fixes are eating into it. It already done been ate. We shouldn't prioritize new work over past work that was never finished. With monthly releases and breaking things down into much smaller chunks it means you have the option to let new work slip to accommodate without moving tasks between people. Ariel On Fri, Apr 10, 2015 at 7:07 PM, Benedict Elliott Smith belliottsm...@datastax.com wrote: CASSANDRA-8459 https://issues.apache.org/jira/browse/CASSANDRA-8459 autocompaction on reads can prevent memtable space reclaimation Can you link a ticket to CASSANDRA-9012 and characterize in a way we can try and implement how to make sufficiently large partitions, over sufficiently large periods of time? Maybe also enumerate the other permutations where this matters like secondary indexes and the access patterns (scans). Does this really qualify for its own ticket? This should just be one of many configurations for stress' part in the new tests. We should perhaps have an aggregation ticket where we ensure we enumerate the configuration data points we've met that need to be covered. But, IMO at least, a methodical exhaustive approach should be undertaken separately, and only be corroborated against such a list to ensure it was done sufficiently well. CASSANDRA-8619 https://issues.apache.org/jira/browse/CASSANDRA-8619 - using CQLSSTableWriter gives ConcurrentModificationException OK. I don't think the original fix meets our new definition of done since the was insufficient coverage, and in this case no regression test. To be done you would have to either implement the coverage or