MAJOR BUG WebLogic 4.5.1 and wl-proxy for Netscape server

Question

Hi there. We are having a major problem with our WebLogic cluster that
          cannot be explained, but is bad enough where we can't deploy our
          application and miss our schedule. Here's the setup:
          We have 8 WebLogic servers (4.5.1) running on 2 Sun 5500 servers with
          2GB of RAM each running Solaris 2.6. The web server in front of the
          cluster is Netscape Enterprise Server version 3.6.2. We are using
          SunSoft's 1.1.7_07 (native threads) as our JDK to start the WebLogic
          servers. We are using in-memory replication on this cluster.
          Here's the problem. The first time we start up the cluster and the
          netscape server, things work fine for a while. It seems that everything
          works fine for the first 10 minutes. But then things start to slow
          down and response time grows exponentially until the browser start
          timing out and nothing works. When this happens, I can ping the servers
          directly and they all claim to be up. I can use the WebLogic console
          utility to connect to the cluster and see all 8 servers just sitting
          there, but I cannot access them through the wl-proxy on the netscape
          server. In fact, I can connect directly to one of the servers at port
          7001 with my browser and everything works fine...
          All of this had led me to believe the there is a serious bug in the
          wl-proxy (libproxy.so) for Solaris. The logs do not seem to indicate
          any problems, but this problem happens every time and we are running out
          of options. We have a WebLogic consultant on site, but he has not been
          able to shed any light on this issue.
          Anyone else experience something like this?? thanks in advance.
          --Vinny
          

Answer 1

Hi Prasad.. We do have a support issue open on this. I'll email you directly with the
          case#. We also have a WebLogic consultant onsite working with us, but we're just not
          getting anywhere with this issue and we're mere days away from deployment.. I was just
          hoping someone else in the newsgroups was also facing this same issue. We can recreate it
          every time we start our servers.
          Thanks
          --Vinny
          Prasad Peddada wrote:
          > Vinny,
          >
          > I am filing an issue about this. I would also suggest you call support to follow up
          > on this.
          > 1-888-232-7878 is the support number or [email protected]. I will try my best to
          > also keep track of this.
          >
          > Thanks
          > Prasad
          >
          > Vinny Carpenter wrote:
          >
          > > Hi Prasad.. Unfortunately that is not going to be acceptable. Our deployment date
          > > is November 1 and if we don't get this issue resolved by Thursday, we cannot go
          > > live.
          > >
          > > --Vinny
          > >
          > > Prasad Peddada wrote:
          > >
          > > > Vinny,
          > > > I can get to this but it won't be until next week. I have to install NES on
          > > > solaris box. I hope that's ok. If I get a chance I will try to test it before
          > > > that.
          > > >
          > > > Thanks
          > > > Prasad
          > > >
          > > > Vinny Carpenter wrote:
          > > >
          > > > > Hi Prasad.. We're a big Solaris/Sun shop and so it's going to be difficult to
          > > > > get a NT server and install NES. I can certainly try, but I am not sure we're
          > > > > going to be able to do that.
          > > > >
          > > > > Any other information I can provide?? Thanks
          > > > >
          > > > > --Vinny
          > > > >
          > > > > Prasad Peddada wrote:
          > > > >
          > > > > > Vinny,
          > > > > >
          > > > > > I don't know if you can do this but to further isolate the problem,
          > > > > > would it be possible to try with NES on NT and see if you still have the
          > > > > > problem.
          > > > > >
          > > > > > Thanks
          > > > > > Prasad
          > > > > >
          > > > > > Vinny Carpenter wrote:
          > > > > >
          > > > > > > Hi there. We are having a major problem with our WebLogic cluster that
          > > > > > > cannot be explained, but is bad enough where we can't deploy our
          > > > > > > application and miss our schedule. Here's the setup:
          > > > > > >
          > > > > > > We have 8 WebLogic servers (4.5.1) running on 2 Sun 5500 servers with
          > > > > > > 2GB of RAM each running Solaris 2.6. The web server in front of the
          > > > > > > cluster is Netscape Enterprise Server version 3.6.2. We are using
          > > > > > > SunSoft's 1.1.7_07 (native threads) as our JDK to start the WebLogic
          > > > > > > servers. We are using in-memory replication on this cluster.
          > > > > > >
          > > > > > > Here's the problem. The first time we start up the cluster and the
          > > > > > > netscape server, things work fine for a while. It seems that everything
          > > > > > > works fine for the first 10 minutes. But then things start to slow
          > > > > > > down and response time grows exponentially until the browser start
          > > > > > > timing out and nothing works. When this happens, I can ping the servers
          > > > > > > directly and they all claim to be up. I can use the WebLogic console
          > > > > > > utility to connect to the cluster and see all 8 servers just sitting
          > > > > > > there, but I cannot access them through the wl-proxy on the netscape
          > > > > > > server. In fact, I can connect directly to one of the servers at port
          > > > > > > 7001 with my browser and everything works fine...
          > > > > > >
          > > > > > > All of this had led me to believe the there is a serious bug in the
          > > > > > > wl-proxy (libproxy.so) for Solaris. The logs do not seem to indicate
          > > > > > > any problems, but this problem happens every time and we are running out
          > > > > > > of options. We have a WebLogic consultant on site, but he has not been
          > > > > > > able to shed any light on this issue.
          > > > > > >
          > > > > > > Anyone else experience something like this?? thanks in advance.
          > > > > > >
          > > > > > > --Vinny