Re: [Ganglia-general] Gmetad-to-Gmetad connection
Hi Vladimir, I changed to “scalable on”. I didn’t help. What I see is only the common remote grid view: CPUs Total: 120 Hosts up: 16 Hosts down:2 == Current Load Avg (15, 5, 1m): 3%, 3%, 3% Avg Utilization (last hour): 4% Localtime: 2015-03-25 10:27 === I can’t see any clusters and hosts inside this grid. By netstat I can see that the second Gmetad instance on machine2 periodically connects to the machine1:8651. I don’t see any connections to machine1:8652. The second Gmetad instance has the same ports, but it’s on another machine. Did you mean that it can affect the polling process? Any ideas? Thanks! Sergey On Mar 24, 2015, at 6:51 PM, Vladimir Vuksan vli...@veus.hr wrote: Hi Sergey, Try setting scalable on in gmetad.conf of the second instance. From the stock gmetad.conf # Scalability mode. If on, we summarize over downstream grids, and respect # authority tags. If off, we take on 2.5.0-era behavior: we do not wrap our output # in GRID/GRID tags, we ignore all GRID tags we see, and always assume # we are the authority on data source feeds. This approach does not scale to # large groups of clusters, but is provided for backwards compatibility. # default: on # scalable off I have not used this feature in a long time so not sure how well it scales however it's worth a shot. Does second instance have different interactive and xml ports ? Vladimir On 03/24/2015 09:24 PM, Sergey wrote: I have one Gmetad instance collecting metrics from several clusters of hosts. Then the second Gmetad instance has to pool all data via port 8651 from the first instance and store everything in local RRDS. I can get all data from the second machine via “#nc machine1 8651”, but when I check RRDS, I don’t see any clusters, only Summary_Data folder. Why Gmetad doesn’t write data into RRDS? -- Dive into the World of Parallel Programming The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/___ Ganglia-general mailing list Ganglia-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ganglia-general
Re: [Ganglia-general] Gmetad-to-Gmetad connection
I changed the second Gmetad to scalable off” and it works! Thank you! Sergey On Mar 25, 2015, at 1:48 PM, Vladimir Vuksan vladi...@vuksan.com wrote: I might have misspoke try scalable off. On March 25, 2015 4:26:55 PM EDT, Sergey svin...@apple.com wrote: Hi Vladimir, I changed to “scalable on”. I didn’t help. What I see is only the common remote grid view: CPUs Total: 120 Hosts up: 16 Hosts down:2 == Current Load Avg (15, 5, 1m): nb sp;3%, 3%, 3% Avg Utilization (last hour): 4% Localtime: 2015-03-25 10:27 === I can’t see any clusters and hosts inside this grid. By netstat I can see that the second Gmetad instance on machine2 periodically connects to the machine1:8651. I don’t see any connections to machine1:8652. The second Gmetad instance has the same ports, but it’s on another machine. Did you mean that it can affect the polling process? Any ideas? Thanks! Sergey On Mar 24, 2015, at 6:51 PM, Vladimir Vuksan vli...@veus.hr mailto:vli...@veus.hr wrote: Hi Sergey, Try setting scalable on in gmetad.conf of the second instance. From the stock gmetad.conf # Scalability mode. If on, we summarize over downstream grids, and respect # authority tags. If off, we take on 2.5.0-era behavior: we do not wrap our output # in GRID/GRID tags, we ignore all GRID tags we see, and always assume # we are the authority on d ata source feeds. This approach does not scale to # large groups of clusters, but is provided for backwards compatibility. # default: on # scalable off I have not used this feature in a long time so not sure how well it scales however it's worth a shot. Does second instance have different interactive and xml ports ? Vladimir On 03/24/2015 09:24 PM, Sergey wrote: I have one Gmetad instance collecting metrics from several clusters of hosts. Then the second Gmetad instance has to pool all data via port 8651 from the first instance and store everything in local RRDS. I can get all data from the second machine via “#nc machine1 8651”, but when I check RRDS, I don’t see any clusters, only Summary_Data folder. Why Gmetad doesn’t wr ite data into RRDS? -- Vladimir -- Dive into the World of Parallel Programming The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/___ Ganglia-general mailing list Ganglia-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ganglia-general
Re: [Ganglia-general] Gmetad-to-Gmetad connection
Hi Sergey, Try setting scalable on in gmetad.conf of the second instance. From the stock gmetad.conf # Scalability mode. If on, we summarize over downstream grids, and respect # authority tags. If off, we take on 2.5.0-era behavior: we do not wrap our output # in GRID/GRID tags, we ignore all GRID tags we see, and always assume # we are the authority on data source feeds. This approach does not scale to # large groups of clusters, but is provided for backwards compatibility. # default: on # scalable off I have not used this feature in a long time so not sure how well it scales however it's worth a shot. Does second instance have different interactive and xml ports ? Vladimir On 03/24/2015 09:24 PM, Sergey wrote: I have one Gmetad instance collecting metrics from several clusters of hosts. Then the second Gmetad instance has to pool all data via port 8651 from the first instance and store everything in local RRDS. I can get all data from the second machine via “#nc machine1 8651”, but when I check RRDS, I don’t see any clusters, only Summary_Data folder. Why Gmetad doesn’t write data into RRDS? -- Dive into the World of Parallel Programming The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Ganglia-general mailing list Ganglia-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ganglia-general