Can tomcat serve MPI (parallel) applications?
Hello, I am looking for a solution for a tomcat container, which is supposed to serve a web application, which is using MPI (openmpi) internally. (The servlet is making JNI calls to C library functions. I have validated that this Java-MPI connection runs without problems when NOT using tomcat.) In catalina.sh, I have changed the lines which actually starts tomcat eval "\"$_RUNJAVA\"" "\"$LOGGING_CONFIG\"" $LOGGING_MANAGER ... into the same command preceded by mpirun: eval mpirun -n 2 "\"$_RUNJAVA\"" "\"$LOGGING_CONFIG\"" $LOGGING_MANAGER ... However, in catalina.out I get errors like "... java.net.BindException: Address already in use ..." This makes sense, because both MPI processes will try to bind to the same address. Is there any chance to have tomcat serve a web application which is using native MPI functions inside one of its servlets? I am using apache-tomcat-7.0.65 on Ubuntu 15.04. Thank you Martijn
Re: Today's Tomcat 9 HTTP/2 webinar is now available on YouTube
> All, > > As promised, today's webinar "Apache Tomcat 9: HTTP/2 Quick Start" is > now available on the Apache Tomcat YouTube channel: > > https://www.youtube.com/channel/UCpqpJ0-G1lYfUBQ6_36Au_g > > Mark > We run all our tomcats behind reverse proxies. Would be interesting how server push could work in such an environment. Regards, Christoph This Email was scanned by Sophos Anti Virus
Re: Today's Tomcat 9 HTTP/2 webinar is now available on YouTube
On 26/11/2015 09:07, Christoph Nenning wrote: >> All, >> >> As promised, today's webinar "Apache Tomcat 9: HTTP/2 Quick Start" is >> now available on the Apache Tomcat YouTube channel: >> >> https://www.youtube.com/channel/UCpqpJ0-G1lYfUBQ6_36Au_g >> >> Mark >> > > > We run all our tomcats behind reverse proxies. Would be interesting how > server push could work in such an environment. You'd need an HTTP/2 compliant reverse proxy. The proxy would then decide whether to push the resource to the client, cache it on the proxy or just drop it on the floor. How the proxy decides what to do is TBD. Mark - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: Today's Tomcat 9 HTTP/2 webinar is now available on YouTube
> >> All, > >> > >> As promised, today's webinar "Apache Tomcat 9: HTTP/2 Quick Start" is > >> now available on the Apache Tomcat YouTube channel: > >> > >> https://www.youtube.com/channel/UCpqpJ0-G1lYfUBQ6_36Au_g > >> > >> Mark > >> > > > > > > We run all our tomcats behind reverse proxies. Would be interesting how > > server push could work in such an environment. > > You'd need an HTTP/2 compliant reverse proxy. The proxy would then > decide whether to push the resource to the client, cache it on the proxy > or just drop it on the floor. How the proxy decides what to do is TBD. > > Mark > > Well, in our case the proxy does not have all the resources to be pushed and not the knowledge which resources should be pushed. For us it would be great if the application could push through the proxy to the client. Regards, Christoph This Email was scanned by Sophos Anti Virus
Re: Today's Tomcat 9 HTTP/2 webinar is now available on YouTube
On 26/11/2015 09:46, Christoph Nenning wrote: All, As promised, today's webinar "Apache Tomcat 9: HTTP/2 Quick Start" is now available on the Apache Tomcat YouTube channel: https://www.youtube.com/channel/UCpqpJ0-G1lYfUBQ6_36Au_g Mark >>> >>> >>> We run all our tomcats behind reverse proxies. Would be interesting > how >>> server push could work in such an environment. >> >> You'd need an HTTP/2 compliant reverse proxy. The proxy would then >> decide whether to push the resource to the client, cache it on the proxy >> or just drop it on the floor. How the proxy decides what to do is TBD. >> >> Mark >> >> > > Well, in our case the proxy does not have all the resources to be pushed > and not the knowledge which resources should be pushed. For us it would be > great if the application could push through the proxy to the client. Sorry if I wasn't clear. The app would push to the proxy and then the proxy decides whether to push to client, cache or drop the request. Mark - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: Today's Tomcat 9 HTTP/2 webinar is now available on YouTube
> All, > > As promised, today's webinar "Apache Tomcat 9: HTTP/2 Quick Start" is > now available on the Apache Tomcat YouTube channel: > > https://www.youtube.com/channel/UCpqpJ0-G1lYfUBQ6_36Au_g > > Mark > > >>> > >>> > >>> We run all our tomcats behind reverse proxies. Would be interesting > > how > >>> server push could work in such an environment. > >> > >> You'd need an HTTP/2 compliant reverse proxy. The proxy would then > >> decide whether to push the resource to the client, cache it on the proxy > >> or just drop it on the floor. How the proxy decides what to do is TBD. > >> > >> Mark > >> > >> > > > > Well, in our case the proxy does not have all the resources to be pushed > > and not the knowledge which resources should be pushed. For us it would be > > great if the application could push through the proxy to the client. > > Sorry if I wasn't clear. The app would push to the proxy and then the > proxy decides whether to push to client, cache or drop the request. > > Mark > Alright, we just need to patch mod_proxy :) Regards, Christoph This Email was scanned by Sophos Anti Virus
Re: Fwd:
Thanks a lot Christopher, my response is below. On Tue, Nov 24, 2015 at 2:26 PM, Christopher Schultz < ch...@christopherschultz.net> wrote: > Yuval, > > On 11/23/15 8:38 AM, Yuval Schwartz wrote: > > I am using tomcat 8.0.22.0 > > Language: Java > > development: Windows, NetBeans. > > deployment: ec2 linux > > > > I am still having trouble understanding/working with "path" and "docBase" > > of context element. > > > > I have two questions: > > > > First of all, I have a context.xml file in the WEB-INF folder of my > > application (call it "myApplication" henceforth) that looks like: > > > > > > You should never specify the "path" attribute in your context.xml file. > Instead, name the WAR file (or exploded directory) the same name as you > want your context's path to be. For a path of "" (empty), use the name > ROOT.war (or ROOT directory). > Executed. However, when I work in my development/local environment (not sure what the correct terminology is for this) I still specify path="" (since, when deploying with netbeans, a file is not generated in the tomcat/webapps directory). Is there a preferred way to do this on my local environment? When deployed on a Linux server, Is it ok to modify the context.xml file in tomcat/webapps/myApplication/META-INF (I think tomcat automcatically redeployed after I made a change)? > > > > maxActive="100" maxIdle="30" maxWait="1" > > username="blah" password="blahblah" > > driverClassName="com.mysql.jdbc.Driver" > > url="jdbc:mysql://localhost:3306/some_db?autoReconnect=true" > > logAbandoned="true" removeAbandoned="true" > > removeAbandonedTimeout="60" type="javax.sql.DataSource" /> > > > > > > My context path is set to the empty string because once I set up a domain > > name it will be redundant to have both the application name and domain > name > > in the url (they are the same). > > See above. > > > This works on my development environment. > > You may notice that your application is deployed on both context paths, > taking up twice as much heap space. Or it might not. Using "path" within > a context.xml file not well-defined, but it is expressly prohibited in > the documentation[1] though not actively prohibited by code. > > > When I deploy this to a linux server of an ec2 instance (on aws) I add a > > line to the server.xml file (in tomcat/conf) under the element > which > > reads: > > > > You should not do this. Instead, drop your ROOT.war into Tomcat's > deployment directory. If you move to EBS, you'll just be deploying the WAR file anyway and shouldn't touch the application servers at all. > Right, I did this and it worked. When you say "EBS" do you mean Amazon's Elastic Beanstalk? or Amazon's Elastic Block Store? > > > (If I don't add this element to the server.xml file then entering just > the > > domain in the url takes me to the Apache Tomcat "successful installation" > > page rather than to the welcome page of myApplication). > > Use ROOT.war. > > > First question: > > Is this the correct way to set up tomcat so that my application has no > > context path? > > No. Note that it /does/ have a context path: it's just an empty string. > > > Not sure if this leads to the "double deployment" problem or > > not. When I open the tomcat manager I see two applications: one with > path: > > "/" and the other with path "/myApplication". (for some reason, I don't > > have the option to "undeploy" the application with path: "/") > > You don't have that option because you have put your into > server.xml. > Removed Context element from Server.xml. > > > Second question: > > The resource element within the context element of my context.xml file > > (located in WEB-INF) specifies a url that is specific to my development > (ie > > local) environment. > > Which URL? The JDBC URL? > The URL attribute of the Resource element, specified above as: url="jdbc:mysql://localhost:3306/some_db?autoReconnect=true" But I think you can disregard the sentence above. I'm not even sure that once I'll deploy I'll want to connect to my development DB instead of my deployed DB (ie: maybe I'll want to connect to the deployed DB in both cases...) > > > When I deploy myApplication to the ec2 instance I would like the resource > > element to have a different url attribute (since it will now be > connecting > > with a database on an aws server). > > One way that I found to do this was to add to the context.xml file in the > > tomcat/conf directory of my server. > > Nonnononoon. Never modify the conf/context.xml file... that changes the > defaults for every application deployed. While you may be deploying only > a single application, it then separates your configuration into two > files when one single file will do. > > > There I add a resource element (between > > the existing context element) which is identical to the one above only > with > > a url attribute that points to the database hosted on aws. > > Package your WAR file
Re: Fwd:
On Tue, Nov 24, 2015 at 3:35 PM, Caldarale, Charles R < chuck.caldar...@unisys.com> wrote: > > From: Christopher Schultz [mailto:ch...@christopherschultz.net] > > Subject: Re: Fwd: > > > > First of all, I have a context.xml file in the WEB-INF folder of my > > > application (call it "myApplication" henceforth) that looks like: > This was a typo, I meant META-INF. Thanks > > Also note that WEB-INF is not the correct directory for context.xml. The > OP really, really needs to read the documentation: > http://tomcat.apache.org/tomcat-8.0-doc/config/context.html > > The OP also needs to learn to use a subject line - he has a history of > failing to do so. > > - Chuck > > > - > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > For additional commands, e-mail: users-h...@tomcat.apache.org > >
Re: [ANN] Apache Tomcat 8.0.29 available
Violeta, On 11/26/15 2:09 AM, Violeta Georgieva wrote: > Chris, > > 2015-11-26 2:52 GMT+02:00 Christopher Schultz> : >> >> Violeta, >> >> On 11/25/15 3:00 PM, Violeta Georgieva wrote: >>> 2015-11-25 21:38 GMT+02:00 Felix Schumacher < >>> felix.schumac...@internetallee.de>: Am 25. November 2015 20:24:17 MEZ, schrieb Violeta Georgieva < >>> miles...@gmail.com>: > Hi, > > 2015-11-25 20:42 GMT+02:00 David Balažic : >> >> Typo on http://tomcat.apache.org/tomcat-8.0-doc/changelog.html >> >> "TLSv1.0 is no an alias" >> >> Should probably be "TLSv1.0 is not an alias" > > I fixed it. I believe it should have been "TLSv1.0 is no(w) an alias ... and will >>> no(t) work ..." >>> >>> Yep you are right. >>> I fixed it. >> >> Also, I'm fairly sure: >> >>> Synchronize OpenSSL to JSSE cipher mapping to recent OpenSSL >>> changes. In particular, TLSv1.0 is no an alias for those ciphers that >>> require TLSv1 and will no work with SDSLv3. >> >> s/SDSLv3/SSLv3/ > > I fixed that with the first commit. Do I miss something? Sorry, I only read this thread... I hadn't seen the commits. -chris - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
SSL Session Reuse in APR based connector
Hi, we are currently running tomcat 8 and I am trying to achieve higher performance. one of the process is to use the SSL Session reuse which will reduce the CPU intensive computation. Can someone let me know if its supported for APR based connector and also let me know the right parameter to use? Regards, Sanaullah
Re: SSL Session Reuse in APR based connector
Sanullah, On 11/26/15 11:21 AM, Sanaullah wrote: > we are currently running tomcat 8 and I am trying to achieve higher > performance. one of the process is to use the SSL Session reuse which will > reduce the CPU intensive computation. > > Can someone let me know if its supported for APR based connector and also > let me know the right parameter to use? Are you talking about "session tickets"? I don't believe Tomcat supports session tickets using any SSL connector. -chris - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: Tomcat catalina.out log is not rolling using log4j DailyRollingFileAppender
Joleen, On 11/26/15 12:12 AM, Joleen Barker wrote: > Alas, no luck. This is what I found in my directory: > > -rw-r--r--. root root 30694 Nov 25 22:49 catalina > -rw-r--r--. root root 0 Nov 25 22:49 catalina.out > -rw-r--r--. root root 0 Nov 25 22:49 host-manager > -rw-r--r--. root root 31909 Nov 25 22:49 localhost > -rw-r--r--. root root 0 Nov 25 22:49 > localhost_access_log.2015-11-25.txt > -rw-r--r--. root root 0 Nov 25 22:49 manager > [root@centos7sys1 logs]# date > Thu Nov 26 00:07:25 EST 2015 So, those two files are growing without bound? Is the date on the server/JVM the same as what you expect? I have some servers in UTC and others in America/New_York and sometimes I get confused when the dates roll-over at times I'm not expecting. What does your log4j configuration look like? -chris > On Wed, Nov 25, 2015 at 7:22 PM, Joleen Barker> wrote: > >> I changed it back to use the log4j setting as I liked it better and the >> boss wants the files to rollover at midnight each night. Now when I startup >> the catalina.out file is empty and the logs appear to be split between the >> catalina file (not catalina.out) and the localhost file. Some of the >> messages are doubled between the two files. >> >> >> We will see what happens now with the swallowOutput=true. I'm so excited >> to see what is there tomorrow morning. lol >> >> -Joleen >> >> On Wed, Nov 25, 2015 at 6:59 PM, Joleen Barker >> wrote: >> >>> Sorry. I all of a sudden noticed the swallowOutput="false" in the >>> cfcc.xml I had. I changed this to be true and now it appears most of the >>> messages are being written to the locahost file. I don't see any messages >>> missing. I will leave it run over night and see what happens and report >>> tomorrow. >>> >>> -Joleen >>> >>> On Wed, Nov 25, 2015 at 3:50 PM, Joleen Barker >>> wrote: >>> I didn't get too far. For the option b and setting the swallowOutput=true in right place I need to get some clarification. I was looking up more information about the swallowOutput parm so I understood more. In Christopher's comments he spoke about updating the context xml. This is where I am confused. When I think of the word "context" it has always referring to the vendors web page to log in with and the context set is "cfcc". There is a file named cfcc.xml in the /server/conf/Catalina/localhost directory. Then there is the actual context.xml file located in the /server/conf directory. I am not sure which one I would put the swallowOutput parm in. Maybe it would be both. The only items in the context.xml file that are not commented out are the following settings and it's a pretty small file: WEB-INF/web.xml ${catalina.base}/conf/web.xml ... ... ... ... >>> classname="org.apache.catalina.valves.CometConnectionManagerValve" /> The cfcc.xml file that I have has the following: >>> charsetMapperClass="org.apache.catalina.util.CharsetMapper" className="org.apache.catalina.core.StandardContext" cookies="true" crossContext="false" debug="0" displayName="Secure Internet File Transfer Web Services" docBase="/opt/mftcc730/server/webapps/cfcc" mapperClass="org.apache.catalina.core.StandardContextMapper" path="/cfcc" privileged="false" reloadable="false" swallowOutput="false" useHttpOnly="false" useNaming="true" wrapperClass="org.apache.catalina.core.StandardWrapper"> >>> directory="/opt/mftcc730/server/logs" prefix="localhost_cfcc_" suffix=".txt" timestamp="true" verbosity="2"/> >>> connectionName="cfcc" connectionPassword="TgPGKAy//0gDOq2Co5UnM2AE8pM=" connectionURL="jdbc:mysql:// 192.168.1.7:3306/mft730?characterEncoding=UTF8" debug="0" digest="SHA" driverName="com.mysql.jdbc.Driver" roleNameCol="RoleID" userCredCol="Passwd" userNameCol="UserID" userRoleTable="UsersMap" userTable="Users" validate="true"/> >>> type="javax.sql.DataSource"/> Thank you. -Joleen On Wed, Nov 25, 2015 at 10:16 AM, Joleen Barker wrote: > Thank you for some direction. > > I'll go ahead and put back the logging as it was from the vendor using > the logging.properties file, etc. etc. so as to minimize their lack of > support due to me changing a lot in their product. > > Then I'll go on to try option b. > > I'll report back with how it goes. > > -Joleeb > Joleen, > > On 11/24/15 4:31 PM, Joleen Barker wrote: >> I have setup the logrotate using cron in the past and it was very >> successful on the Linux boxes but I could not find an equivalent > setup for >> AIX. Things seem so much easier on Linux. The company wants a >