It doesn't look like this could be causing the problem, but I do see one tiny issue: The XPath you're using could very well get you some false positives (if the root id is 1234 and a node in another node's path has the id 21234)...

A safer XPath would be something like:

//Station[ancestor::*[@id={0}]]

2 cents... /Chriztian

Stephan,

Cool will try this I have updated the content finder also added logging see if i can determine anything from the logs. Although on my dev it all works I dont see anything in umbraco logs which is a bit weird my class now looks like:

public class StationsContentFinder : ContentFinderByNiceUrl
{
    const string StationsXPath = "//Stations[contains(@path,'{0}')]";

    public override bool TryFindContent(PublishedContentRequest request)
    {
        if (base.TryFindContent(request)) return true;

        try
        {
            if (request.HasDomain)
            {
                var rootId = request.UmbracoDomain.RootContentId; // multilingual site we need root 

                Log("working with root id " + rootId);

                Log("using base uri " + request.Uri.AbsolutePath);

                var urlChunks =
                    request.Uri.AbsolutePath.Split("/".ToCharArray(), StringSplitOptions.RemoveEmptyEntries)
                        .ToList();

                if (IsSingleLetter(urlChunks))
                {
                    Log("got letter " + urlChunks.Last().ToUpper());

                    request.RoutingContext.UmbracoContext.HttpContext.Items.Add("StationLetter",
                        urlChunks.Last().ToUpper());

                    string stationXPath = string.Format(StationsXPath, rootId);

                    var stationsNode =
                        request.RoutingContext.UmbracoContext.ContentCache.GetByXPath(stationXPath).FirstOrDefault();

                    if (stationsNode == null)
                    {
                        return false;
                    }

                    Log("found stations node " + stationsNode.Name + " id " + stationsNode);

                    request.PublishedContent = stationsNode;

                    //the stations hijack will take over and test for the letter and get the list of stations

                    return true;
                }
            }
            else
            {
                Log("no domain");
            }
        }
        catch (Exception ex)
        {
            LogHelper.Error<Exception>("Error with station content finder",ex);
        }
        return false;
    }

    private static bool IsSingleLetter(List<string> urlChunks)
    {
        return urlChunks.Last().Length == 1 && Char.IsLetter(urlChunks.Last().ToCharArray()[0]);
    }

    private void Log(string message)
    {
        LogHelper.Debug(typeof(StationsContentFinder), message);
    }
}

Regards

Ismail

Looks good. Point is, in dev you are requesting "known" urls that do match to your domains, whereas in production you are potentially getting "rogue" urls that maybe don't match any domain. Let me know if it works!

Stephan,

Cool will get this deployed and test. Also figured out my logging error i was doing debug but log4net was set to info have set it to all i can now see logs. So hopefully if it goes down again on live we may be able to log urls then see whats going on.

Regards

Ismail

Stephan,

We deployed the update yesterday and so far it seems to be stable. Currently being monitored the client has turned off hourly app pool recycle so fingers crossed.

Thanks for your help.

Regards

Ismail

is working on a reply...