Appreciate your feedback, Mark. I'll get it incorporated ASAP. On Wed, Dec 12, 2018 at 8:26 AM Mark Thomas <ma...@apache.org> wrote:
> On 11/12/2018 19:20, Mallory Mooney wrote: > > Hi all! > > > > It's been a long time coming, so I wanted to resurrect this thread one > more > > time to share the now published post. You can find it at > > https://www.datadoghq.com/blog/tomcat-architecture-and-performance/. > > Now I can see the document without having to authenticate... > > The first architecture diagram: > - doesn't show any hosts > - incorrectly classes a "Service" as a "webapp" (a context is a webapp) > - shows the unusual setup of multiple services > > "Server" and "Service" are not peers. "Service" must be nested inside > "Server". > > The following sentence is misleading at best: > "The engine returns the appropriate response (e.g., JSP page, servlet) > to the client based on the Host request-header field in the request." > > AJP allows integration with multiple reverse proxies. Not just httpd. > > It is extremely debatable whether using httpd for static content is more > efficient. > > The description of the differences between APR/native and NIO[2] is > wrong. All the connectors support non-blocking. When the switch between > non-blocking and simulated blocking happens depends on whether there is > an active request and the APIs used by the application. > > Note there can be more than one acceptor thread. > > The default for minSpareThreads does not agree with the Tomcat docs - it > should be 10. It is 25 for an Executor. > > Using an Executor doesn't "evenly distribute threads across multiple > connectors". By providing a single thread pool for multiple connectors > threads can be used more efficiently. That means that - rather than an > even distribution - a connector could be starved of threads during a > period of heavy load if all the threads are allocated to a different > connector (unlikely but possible). Any admin/monitoring services exposed > via HTTP probably needs to have a dedicated thread pool to ensure it is > always available. > > It would be worth including an upfront warning of about the (lack of) > security in JMX. There is almost no granularity in the access controls > and brute force are easy to mount but hard to detect. The Tomcat > security guide discusses this in more detail. > > maxTime isn't a "cumulative count". It is the longest time for any > single request to complete. > > Tomcat won't return 503s if it runs out of threads. It will simply drop > the connection. If a reverse proxy is being used, it will probably > return a 503 in those circumstances. > > Nothing jumped out at me while I was reading the metrics sections. I'd > probably add a note that an OOME means you need to restart the server. > An OOME might be recoverable or might not. As a user there is no way to > tell. The only safe thing to do is to restart. > > Mark > > > > > I hope everyone has a great week! > > Mallory > > > > > > > > On Sat, Nov 3, 2018 at 4:28 AM Mark Thomas <ma...@apache.org> wrote: > > > >> On 02/11/2018 16:34, Mallory Mooney wrote: > >>> Hi everyone! > >>> > >>> Just wanted to check-in to see if anyone has had the chance to take a > >> look > >>> at the draft and leave feedback. I'll be locking things down for this > one > >>> by Thursday of next week (Nov. 8), so there's some time left if you are > >>> still working through it. > >> > >> Note I suspect a number of folks aren't even looking at it because it is > >> behind an authentication-wall (regardless of how willingly / speedily > >> access is granted on request). > >> > >> Mark > >> > >> > >>> > >>> Have a great weekend! > >>> > >>> > >>> On Fri, Oct 19, 2018 at 7:52 AM Mallory Mooney < > >> mallory.moo...@datadoghq.com> > >>> wrote: > >>> > >>>> Hi all! > >>>> > >>>> Just wanted to give you an update. I am working to get things for this > >>>> series wrapped up by the first week of November. Hopefully, that still > >>>> gives everyone enough time to review and leave any feedback. > >>>> > >>>> Mallory > >>>> > >>>> On Fri, Oct 12, 2018 at 3:37 PM Mallory Mooney < > >>>> mallory.moo...@datadoghq.com> wrote: > >>>> > >>>>> Igal, it will be available publicly once published! I don't have an > >>>>> official publish date yet but can share that when it becomes more > >> concrete. > >>>>> > >>>>> And no PRs yet, Chris! It's still in the less cool GDoc stage of the > >>>>> review process. :) > >>>>> > >>>>> On Fri, Oct 12, 2018 at 1:36 PM Igal Sapir <isa...@apache.org> > wrote: > >>>>> > >>>>>> On 10/12/2018 11:23 AM, Christopher Schultz wrote: > >>> Mallory, > >>> > >>> On 10/12/18 13:23, Mallory Mooney wrote: > >>>>>>>>> I definitely appreciate everyone's willingness to help out! > >>>>>>>>> > >>>>>>>>> Here is the link to the GDoc: > >>>>>>>>> > >>>>>>> > >> > https://docs.google.com/document/d/1fudlXj055nnPd-1lUoAXIS2ge8qNI56_jgUhHgKczFE/edit?usp=sharing > >>>>>>>>> Requesting access will still be needed, but I can grant that > >> ASAP. > >>>>>>>>> I want to make sure I can attribute comments/suggestions to > >>>>>>>>> specific people, so I know who to thank, and who to follow up > with > >>>>>>>>> if I need more clarification on a specific comment. Plus, it > makes > >>>>>>>>> the IT department happy. > >>> What, no GitHub PRs? ;) > >>>>>>> > >>>>>>> +1 > >>>>>>> > >>>>>>> Is that going to be publicly available or is it a > >>>>>>> proprietary/private/commercial guide? > >>>>>>> > >>>>>>> > >>>>>>> Igal > >>>>>>> > >>> -chris > >>> > >>>>>>>>> On Thu, Oct 11, 2018 at 9:53 AM Christopher Schultz < > >>>>>>>>> ch...@christopherschultz.net> wrote: > >>>>>>>>> > >>>>>>>>> Mark, > >>>>>>>>> > >>>>>>>>> On 10/10/18 6:00 PM, Mark Thomas wrote: > >>>>>>>>>>>> On 10/10/18 17:44, Mallory Mooney wrote: > >>>>>>>>>>>>> Hi all, > >>>>>>>>>>>>> > >>>>>>>>>>>>> I work for Datadog and am writing a guide about monitoring > >>>>>>>>>>>>> Tomcat (with or without Datadog). I'd love to get some > >>>>>>>>>>>>> feedback on the technical content. The project maintainers > >>>>>>>>>>>>> we reached out to recommended we post a request here. > >>>>>>>>>>>>> > >>>>>>>>>>>>> Would anyone be up for that? I can send the post link to > >>>>>>>>>>>>> someone directly. > >>>>>>>>>>>>> > >>>>>>>>>>>>> Appreciate your help and time! > >>>>>>>>>>>> Why not post the link here so the community can review the > >>>>>>>>>>>> document? > >>>>>>>>> +1 > >>>>>>>>> > >>>>>>>>> There are many active community members who have an interest in > >>>>>>>>> monitoring. I'm sure you'll get lots of feedback. > >>>>>>>>> > >>>>>>>>> -chris > >>>>>>>>>> > >> --------------------------------------------------------------------- > >>>>>>>>>> > >>>>>>>>>> > >>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > >>>>>>>>>> For additional commands, e-mail: users-h...@tomcat.apache.org > >>>>>>>>>> > >>>>>>>>>> > >>>>>>> > >>>>>>> > --------------------------------------------------------------------- > >>>>>>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > >>>>>>> For additional commands, e-mail: users-h...@tomcat.apache.org > >>>>>>> > >>>>>> > >>>>>> > >>>>>> > --------------------------------------------------------------------- > >>>>>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > >>>>>> For additional commands, e-mail: users-h...@tomcat.apache.org > >>>>>> > >>>>>> > >>>>> > >>>>> -- > >>>>> Mallory Mooney > >>>>> Technical Content Writer > >>>>> > >>>> > >>>> > >>>> -- > >>>> Mallory Mooney > >>>> Technical Content Writer > >>>> > >>> > >>> > >> > >> > >> --------------------------------------------------------------------- > >> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > >> For additional commands, e-mail: users-h...@tomcat.apache.org > >> > >> > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > For additional commands, e-mail: users-h...@tomcat.apache.org > > -- Mallory Mooney Technical Content Writer
