# Exception in thread? **URL:** \ **Category:** ruby-talk\ **Created:** [29 September 2008 16:22 UTC](https://rubytalk.org/t/exception-in-thread/49357 "2008-09-29T16:22:52Z")\ **Posts on this page:** 8\ **Page:** 1

**Author:** ![Sylvain\_Viart1](https://avatars.discourse-cdn.com/v4/letter/s/eb9ed0/32.png) [@Sylvain\_Viart1](https://rubytalk.org/u/Sylvain_Viart1)\ **Post date:** [29 September 2008 16:22 UTC](https://rubytalk.org/t/exception-in-thread/49357/1 "2008-09-29T16:22:52Z")

Hi, I'm trying to use Net::SSH::Multi. [http://net-ssh.rubyforge.org/](http://net-ssh.rubyforge.org/) It seems that ruby have difficulties to catch exception in multi threaded mode, any hint? the doc said, if we put (:on\_error =\> :warn) it shouldn't fail, be the exception begin/rescue bloc did catch the exception, here Errno::EHOSTUNREACH /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/./multi/session.rb 468 def next\_session(server, force=false) #:nodoc: [...] 482 begin 483 server.new\_session 484 rescue Exception =\> e 485 server.fail! 486 @session\_mutex.synchronize { @open\_connections -= 1 } 487 488 case on\_error 489 when :ignore then 490 # do nothing 491 when :warn then 492 warn("error connecting to #{server}: #{e.class} (#{e.message})") 493 when Proc then 494 go = catch(:go) { on\_error.call(server); nil } 495 case go 496 when nil, :ignore then # nothing 497 when :retry then retry 498 when :raise then raise 499 else warn "unknown 'go' command: #{go.inspect}" 500 end 501 else 502 raise 503 end 504 505 return nil 506 end [...] As we can test, we are able to catch the exception at to level, could you confirm its tread related? require 'rubygems' require 'net/ssh/multi' Net::SSH::Multi.start(:on\_error =\> :warn) do |session| # define the servers we want to use session.use 'root@server-04' session.use 'root@server-07' # doesn't exist session.use 'root@server-08' # execute commands on all servers begin session.exec( "hostname" ) rescue Exception =\> e p "main:#{e}" end # run the aggregated event loop session.loop end ruby 1.8.5 (2006-08-25) [x86\_64-linux] Regards, Sylvain. ---

**Author:** ![Brian\_Candler](https://avatars.discourse-cdn.com/v4/letter/b/5f9b8f/32.png) [@Brian\_Candler](https://rubytalk.org/u/Brian_Candler)\ **Post date:** [30 September 2008 08:53 UTC](https://rubytalk.org/t/exception-in-thread/49357/2 "2008-09-30T08:53:22Z")

Sylvain Viart wrote: > It seems that ruby have difficulties to catch exception in multi > threaded mode, any hint? For debugging purposes, maybe you want Thread.abort\_on\_exception = true (or just run ruby with -d flag) Other than that I don't understand your problem. What behaviour do you see when you run your test program? What behaviour do you expect? Is no warning generated for the non-existent host? > # execute commands on all servers > begin > session.exec( "hostname" ) > rescue Exception =\> e > p "main:#{e}" > end That rescue won't catch exceptions in other threads. Each thread of execution is responsible for catching its own exceptions. If it doesn't, then the thread just terminates (unless Thread.abort\_on\_exception is set) It \*is\* possible for one thread to raise an exception in another thread (Thread#raise), but this is extremely hairy asynchronous programming and I would strongly discourage it. It would seem reasonable for session.exec to collect the status of each of the threads and return an array of them. I don't know if it does so. Perhaps you can use something like this: errs = ... :on\_error =\> lambda { |server| errs \<\< server } ... session.exec "hostname" unless errs.empty? puts "The command failed on #{errs.size} hosts" end > **···** > > -- > Posted via [http://www.ruby-forum.com/\](http://www.ruby-forum.com/%5C). ---

**Author:** ![Sylvain\_Viart1](https://avatars.discourse-cdn.com/v4/letter/s/eb9ed0/32.png) [@Sylvain\_Viart1](https://rubytalk.org/u/Sylvain_Viart1)\ **Post date:** [30 September 2008 11:13 UTC](https://rubytalk.org/t/exception-in-thread/49357/3 "2008-09-30T11:13:55Z")

Hi Brian, Brian Candler a écrit : > Sylvain Viart wrote: > > > > It seems that ruby have difficulties to catch exception in multi > > threaded mode, any hint? > > > > For debugging purposes, maybe you want Thread.abort\_on\_exception = true (or just run ruby with -d flag) > > Thanks good to know that. > Other than that I don't understand your problem. What behaviour do you see when you run your test program? What behaviour do you expect? Sorry, I was late yesterday and my post is confusing. In fact, I've made some tests and I suspect some strange behavior (or unknown to me) on exception handling. In the lib, we got a bloc with 484 rescue Exception =\> e Which I would expect to catch anything. but it missed Errno::EHOSTUNREACH. I didn't find a good explanation so I suspect that exception in threads are behaving somewhat differently. Strangely, if I add another rescue statement in the lib: /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/./multi/session.rb 506 rescue 507 puts "caught:#{$!}" 508 end it works?? Why? Why the normally more open "rescue Exception =\> e" didn't do its job? That's why I suspect some interaction between exception and threaded execution. It seems I missed something about exception or so. :-\ > Is no warning generated for the non-existent host? > > > # execute commands on all servers > > begin > > session.exec( "hostname" ) > > rescue Exception =\> e > > p "main:#{e}" > > end > > Sorry for that, I was expecting this bloc not using the begin/rescue in fact. The rescue here, catch the Errno::EHOSTUNREACH., not caught internally by the lib. I should have written: session.exec( "hostname" ) With no rescue, the program fail, no job is performed on any host. > It would seem reasonable for session.exec to collect the status of each of the threads and return an array of them. I don't know if it does so. Perhaps you can use something like this: > > errs = > ... > :on\_error =\> lambda { |server| errs \<\< server } > ... > session.exec "hostname" > unless errs.empty? > puts "The command failed on #{errs.size} hosts" > end > Hum, nice, I'm gonna try. 🙂 Would it catch the Errno::EHOSTUNREACH? Thanks for your hints. Regards, Sylvain. ---

**Author:** ![Brian\_Candler](https://avatars.discourse-cdn.com/v4/letter/b/5f9b8f/32.png) [@Brian\_Candler](https://rubytalk.org/u/Brian_Candler)\ **Post date:** [30 September 2008 14:13 UTC](https://rubytalk.org/t/exception-in-thread/49357/4 "2008-09-30T14:13:54Z")

Sylvain Viart wrote: > In the lib, we got a bloc with > > 484 rescue Exception =\> e > > Which I would expect to catch anything. but it missed > Errno::EHOSTUNREACH. Probably I should not try to answer this as I don't use Net::SSH::Multi, but I've installed the gem now: Successfully installed net-ssh-2.0.4 Successfully installed net-ssh-gateway-1.0.0 Successfully installed net-ssh-multi-1.0.0 I see that rescue only covers the preceding line: begin server.new\_session rescue Exception =\> e That is, it will catch an exception raised by server.new\_session only. > Strangely, if I add another rescue statement in the lib: > > /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/./multi/session.rb > 506 rescue > 507 puts "caught:#{$!}" > 508 end > > it works?? Why? Possibly it's rescuing an exception which is occurring between lines 469 and 480. But since you didn't show the actual backtrace, then this is pure guesswork. One option is: puts "caught:#{$!}\n#{$!.backtrace.join("\n")}" > Why the normally more open "rescue Exception =\> e" didn't do its job? I don't know. But when making a extraordinary claim ("rescue is not doing its job") then you need to provide the evidence to back it up. Now, I can replicate something like your problem: pointing to a non-existent host on my LAN gives an Errno::EHOSTUNREACH. require 'rubygems' require 'net/ssh/multi' Net::SSH::Multi.start(:on\_error =\> :warn) do |session| # define the servers we want to use session.use 'root@localhost' session.use 'root@10.1.1.10' # non-existent host on local LAN # execute commands on all servers begin session.exec( "hostname" ) rescue Exception =\> e puts "main:#{e}\n#{e.backtrace.join("\n")}" end # run the aggregated event loop session.loop end $ ruby test.rb error connecting to root@localhost: Net::SSH::AuthenticationFailed (root@localhost) main:No route to host - connect(2) /usr/local/lib/ruby/gems/1.8/gems/net-ssh-2.0.4/lib/net/ssh/transport/session.rb:65:in `initialize' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `join' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `sessions' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `each' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `sessions' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:81:in `open\_channel' /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:120:in `exec' test.rb:10 /usr/local/lib/ruby/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi.rb:62:in `start' test.rb:3 I get an exception like you do. However I see no evidence at all that lib/net/ssh/multi/session.rb is involved. This is ruby 1.8.6p114. I don't have any specific reason why 1.8.5 wouldn't work, but I consider 1.8.6p114 to be the most "stable" Ruby available (certainly more stable than later releases in the 1.8.6 family 🙂 and it may well be that Net::SSH::Multi hasn't been well tested with 1.8.5. So it could be worth a try. > **···** > > -- > Posted via [http://www.ruby-forum.com/\](http://www.ruby-forum.com/%5C). ---

**Author:** ![Sylvain\_Viart1](https://avatars.discourse-cdn.com/v4/letter/s/eb9ed0/32.png) [@Sylvain\_Viart1](https://rubytalk.org/u/Sylvain_Viart1)\ **Post date:** [30 September 2008 15:36 UTC](https://rubytalk.org/t/exception-in-thread/49357/5 "2008-09-30T15:36:11Z")

Hi, Sylvain Viart a écrit : > > It would seem reasonable for session.exec to collect the status of each of the threads and return an array of them. I don't know if it does so. Perhaps you can use something like this: > > > > errs = > > ... > > :on\_error =\> lambda { |server| errs \<\< server } > > ... > > session.exec "hostname" > > unless errs.empty? > > puts "The command failed on #{errs.size} hosts" > > end > > Hum, nice, I'm gonna try. 🙂 > Would it catch the Errno::EHOSTUNREACH? It didn't catch the exception a workaround, using a closure and Net::SSH::Multi::DynamicServer \<[http://net-ssh.rubyforge.org/multi/v1/api/classes/Net/SSH/Multi/DynamicServer.html>](http://net-ssh.rubyforge.org/multi/v1/api/classes/Net/SSH/Multi/DynamicServer.html>); behavior, instead of specifying the server. It's evaluated by attempting the ssh connection first. Which mean the server, is connected twice, during the test and later in the session. Note that I discard the 'options' to test the connection. errs = def test\_server(errs, server) lambda do |options| begin server =~ /(.+)@(.+)/ server\_name, user = $2, $1 puts server\_name s = Net::SSH.start(server\_name, user) s.close s = server rescue Errno::EHOSTUNREACH, SocketError puts "echec connexion #{server} : #{$!}" errs \<\< server s = nil end return s end end Net::SSH::Multi.start(:on\_error =\> :warn) do |session| # define the servers we want to use session.use &test\_server(errs, 'root@srv-04') session.use &test\_server(errs, 'root@srv-07') session.use &test\_server(errs, 'root@srv-08') session.use &test\_server(errs, 'root@fail-08.local') # execute commands on all servers session.exec( "hostname" ) # run the aggregated event loop session.loop end unless errs.empty? puts "The command failed on #{errs.size} hosts" end #srv-04 #srv-07 #echec connexion root@srv-07 : No route to host - connect(2) #srv-08 #fail-08.local #echec connexion root@fail-08.local : getaddrinfo: Name or service not known #[srv-04] srv-04 #[srv-08] srv-08 #The command failed on 2 hosts Works, but I still don't know why the exception are not handled in the lib Net::SSH::Multi which may be specific to this lib. I still appreciate some more hint. Regards, Sylvain. ---

**Author:** ![Brian\_Candler](https://avatars.discourse-cdn.com/v4/letter/b/5f9b8f/32.png) [@Brian\_Candler](https://rubytalk.org/u/Brian_Candler)\ **Post date:** [30 September 2008 20:49 UTC](https://rubytalk.org/t/exception-in-thread/49357/6 "2008-09-30T20:49:07Z")

Sylvain Viart wrote: > Works, but I still don't know why the exception are not handled in the > lib Net::SSH::Multi which may be specific to this lib. Show the backtrace! Otherwise, nobody is going to be able to help you. That is, in your original demo code, either remove the top-level rescue clause entirely, or change it to rescue Exception =\> e puts "main:#{e}\n#{e.backtrace.join("\n")}" end Then paste the full, unedited result here. > **···** > > -- > Posted via [http://www.ruby-forum.com/\](http://www.ruby-forum.com/%5C). ---

**Author:** ![Sylvain\_Viart1](https://avatars.discourse-cdn.com/v4/letter/s/eb9ed0/32.png) [@Sylvain\_Viart1](https://rubytalk.org/u/Sylvain_Viart1)\ **Post date:** [3 October 2008 07:39 UTC](https://rubytalk.org/t/exception-in-thread/49357/7 "2008-10-03T07:39:55Z")

Hi Brian, Thanks a lot for your work, I really appreciate your effort. 🙂 Brian Candler a écrit : > Sylvain Viart wrote: > > > > Works, but I still don't know why the exception are not handled in the > > lib Net::SSH::Multi which may be specific to this lib. > > > > Show the backtrace! Otherwise, nobody is going to be able to help you. > > Sorry for that, I'm not enough backtrace friendly. :-\ ----------------------------8\<----------------------- t3.rb require 'rubygems' require 'net/ssh/multi' Net::SSH::Multi.start(:on\_error =\> :warn) do |session| # define the servers we want to use session.use 'root@srv-04' session.use 'root@srv-07' session.use 'root@srv-08' session.use 'root@fail-08.local' # execute commands on all servers session.exec( "hostname" ) # run the aggregated event loop session.loop end ----------------------------8\<----------------------- ruby t3.rb error connecting to root@srv-04: Net::SSH::AuthenticationFailed (root@srv-04) Text will be echoed in the clear. Please install the HighLine or Termios libraries to suppress echoed text. Password: /var/lib/gems/1.8/gems/net-ssh-2.0.4/lib/net/ssh/transport/session.rb:65:in `initialize': No route to host - connect(2) (Errno::EHOSTUNREACH) from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `join' from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `sessions' from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `each' from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:37:in `sessions' from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:81:in `open\_channel' from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi/session\_actions.rb:120:in `exec' from t3.rb:12 from /var/lib/gems/1.8/gems/net-ssh-multi-1.0.0/lib/net/ssh/multi.rb:62:in `start' from t3.rb:4 shell returned 1 I think you're right and this kind of exception is not handled by the lib. I've to reread the lib, but its doc is confusing. Regards, Sylvain. ---

**Author:** ![Brian\_Candler](https://avatars.discourse-cdn.com/v4/letter/b/5f9b8f/32.png) [@Brian\_Candler](https://rubytalk.org/u/Brian_Candler)\ **Post date:** [3 October 2008 07:57 UTC](https://rubytalk.org/t/exception-in-thread/49357/8 "2008-10-03T07:57:31Z")

It does seem inconsistent that "error connecting to root@srv-04: Net::SSH::AuthenticationFailed" is caught as a warning, but Errno::EHOSTUNREACH is not. I suggest you check for a project mailing list or bug tracker and report it there. [http://rubyforge.org/projects/net-ssh](http://rubyforge.org/projects/net-ssh) > **···** > > -- > Posted via [http://www.ruby-forum.com/](http://www.ruby-forum.com/).