Forum Replies Created
-
Posts
-
Hi, I tried the provided remedy which lead to the same error “go: command not found”.
However, I did find a possibly temporary solution. Instead of saying “go install ….“, you list the full path rather than the alias itself. For example: “/usr/local/go/bin/go install …“. This solution might be limited as it will not work in situations where programs need a reference to GOPATH or PATH (w Go included).
“*** Use a separate network for each wide-area connection that uses Basic NICs to connect a pair of nodes. Each node may have several Basic NICs connected other nodes. You probably don’t want to use this method if you need to connect all pairs of nodes.”
Yes, this would be ideal. Let me explain.
We initially deployed our application with 5 nodes with 2 sites involved. This is because our application requires 4 nodes and we use one as the client for our application. To be mindful of other projects on FABRIC, we figured our first few deployments, we would use the least amount of resources to test all conditions (see if site to site works and to run our app successfully).
That said, I think our official deployment (that professors/students start to use), we would like to ask you how many sites/nodes would be best to use here. I know our project lead would like to store large amounts of genomes and so this likely requires a future call to plan. Additionally, it would be nice to have many sites to expand the availability of our data that we publish into the application.
Also, we are doing routing manually. Having all the nodes connected together allows us to have a easier time setting things up. As we switch to having a network that’s more realistic, one concern is “link costs” (if that’s something that you have data for and could give us). The cost of a certain link is plugged into our routing protocol (it uses some kind of LCR). I’m am sure you know this, but just as an example: a 100Gbps path would have a lesser cost than a 1Gbps path. Also it is not just about the bandwidth but also about latency. I am sure you have data like this that we could or we could at the very least, we could test and figure out the cost ourselves.
-
This reply was modified 2 years, 8 months ago by
.
-
This reply was modified 2 years, 8 months ago by
-
This reply was modified 2 years, 8 months ago by
I have two of the IPv6 node addresses if you can find the sites (my partner can only access the slice):
- [2001:1948:417:7:f816:3eff:fe94:5413]
- [2001:400:a100:3030:f816:3eff:feef:dba1]
If I find out the sites, I will reply again. To add, we have only tested two sites, so there very well could be more hops that limit packet sizes. We will let you know as we test more! Thank you.
Kind regards,
Justin
It might help to have some more details.
We have a service (NDN) running on top of the IP layer on port 6363.
We want to not only allow easy internal communication but also external which we plan to figure out later. We initially ran into some issues using IPv4 as our packets were not getting through. We figured that our packets were dropped by the public address and not forwarding the packets to the desired private address due to not knowing about this communication.
IPv6 was the ‘goto’ solution as this communication issue does not occur and would be easiest to use.
I am assuming there is something we could set up (in our slice creation) to let the public addresses at the sites we are using to know about / allow the communication?
Kind regards,
Justin
-
This reply was modified 2 years, 8 months ago by
